OVMS3-idf/components/esp_ringbuf/test/test_ringbuf.c

628 lines
26 KiB
C
Raw Normal View History

#include <stdio.h>
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
#include <stdlib.h>
#include "freertos/FreeRTOS.h"
#include "freertos/task.h"
#include "freertos/queue.h"
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
#include "freertos/semphr.h"
#include "freertos/ringbuf.h"
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
#include "driver/timer.h"
#include "esp_spi_flash.h"
#include "unity.h"
#include "test_utils.h"
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Definitions used in multiple test cases
#define TIMEOUT_TICKS 10
#define NO_OF_RB_TYPES 3
#define ITEM_HDR_SIZE 8
#define SMALL_ITEM_SIZE 8
#define LARGE_ITEM_SIZE (2 * SMALL_ITEM_SIZE)
#define BUFFER_SIZE 160 //4Byte aligned size
static const uint8_t small_item[SMALL_ITEM_SIZE] = { 0x00, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07};
static const uint8_t large_item[LARGE_ITEM_SIZE] = { 0x08, 0x09, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, 0x0F,
0x10, 0x11, 0x12, 0x13, 0x14, 0x15, 0x16, 0x17};
static RingbufHandle_t buffer_handles[NO_OF_RB_TYPES];
static SemaphoreHandle_t done_sem;
static void send_item_and_check(RingbufHandle_t handle, const uint8_t *item, size_t item_size, TickType_t ticks_to_wait, bool in_isr)
{
BaseType_t ret;
if (in_isr) {
ret = xRingbufferSendFromISR(handle, (void *)item, item_size, NULL);
} else {
ret = xRingbufferSend(handle, (void *)item, item_size, ticks_to_wait);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_ASSERT_MESSAGE(ret == pdTRUE, "Failed to send item");
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
static void receive_check_and_return_item_no_split(RingbufHandle_t handle, const uint8_t *expected_data, size_t expected_size, TickType_t ticks_to_wait, bool in_isr)
{
//Receive item from no-split buffer
size_t item_size;
uint8_t *item;
if (in_isr) {
item = (uint8_t *)xRingbufferReceiveFromISR(handle, &item_size);
} else {
item = (uint8_t *)xRingbufferReceive(handle, &item_size, ticks_to_wait);
}
TEST_ASSERT_MESSAGE(item != NULL, "Failed to receive item");
TEST_ASSERT_MESSAGE(item_size == expected_size, "Item size is incorrect");
//Check data of received item
for (int i = 0; i < item_size; i++) {
TEST_ASSERT_MESSAGE(item[i] == expected_data[i], "Item data is invalid");
}
//Return item
if (in_isr) {
vRingbufferReturnItemFromISR(handle, (void *)item, NULL);
} else {
vRingbufferReturnItem(handle, (void *)item);
}
}
static void receive_check_and_return_item_allow_split(RingbufHandle_t handle, const uint8_t *expected_data, size_t expected_size, TickType_t ticks_to_wait, bool in_isr)
{
//Receive item
size_t item_size1, item_size2;
uint8_t *item1, *item2;
BaseType_t ret;
if (in_isr) {
ret = xRingbufferReceiveSplitFromISR(handle, (void**)&item1, (void **)&item2, &item_size1, &item_size2);
} else {
ret = xRingbufferReceiveSplit(handle, (void**)&item1, (void **)&item2, &item_size1, &item_size2, ticks_to_wait);
}
//= xRingbufferReceiveSplit(handle, (void**)&item1, (void **)&item2, &item_size1, &item_size2, ticks_to_wait);
TEST_ASSERT_MESSAGE(ret == pdTRUE, "Failed to receive item");
TEST_ASSERT_MESSAGE(item1 != NULL, "Failed to receive item");
//Check data of received item(s) and return them
if (item2 == NULL) {
TEST_ASSERT_MESSAGE(item_size1 == expected_size, "Item size is incorrect");
for (int i = 0; i < item_size1; i++) {
TEST_ASSERT_MESSAGE(item1[i] == expected_data[i], "Item data is invalid");
}
//Return item
if (in_isr) {
vRingbufferReturnItemFromISR(handle, (void *)item1, NULL);
} else {
vRingbufferReturnItem(handle, (void *)item1);
}
} else {
//Item was split
TEST_ASSERT_MESSAGE(item_size1 + item_size2 == expected_size, "Total item size is incorrect");
for (int i = 0; i < item_size1; i++) {
TEST_ASSERT_MESSAGE(item1[i] == expected_data[i], "Head item data is invalid");
}
for (int i = 0; i < item_size2; i++) {
TEST_ASSERT_MESSAGE(item2[i] == expected_data[item_size1 + i], "Head item data is invalid");
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Return Items
if (in_isr) {
vRingbufferReturnItemFromISR(handle, (void *)item1, NULL);
vRingbufferReturnItemFromISR(handle, (void *)item2, NULL);
} else {
vRingbufferReturnItem(handle, (void *)item1);
vRingbufferReturnItem(handle, (void *)item2);
}
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
static void receive_check_and_return_item_byte_buffer(RingbufHandle_t handle, const uint8_t *expected_data, size_t expected_size, TickType_t ticks_to_wait, bool in_isr)
{
//Receive item
size_t item_size;
uint8_t *item;
if (in_isr) {
item = (uint8_t *)xRingbufferReceiveUpToFromISR(handle, &item_size, expected_size);
} else {
item = (uint8_t *)xRingbufferReceiveUpTo(handle, &item_size, ticks_to_wait, expected_size); //Limit amount of bytes returned to the size of one item
}
TEST_ASSERT_MESSAGE(item != NULL, "Failed to receive item");
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Check data of received item
for (int i = 0; i < item_size; i++) {
TEST_ASSERT_MESSAGE(item[i] == expected_data[i], "Item data is invalid");
}
//Return item
if (in_isr) {
vRingbufferReturnItemFromISR(handle, (void *)item, NULL);
} else {
vRingbufferReturnItem(handle, (void *)item);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Check if item wrapped around
if (item_size < expected_size) {
//Item is wrapped, receive second portion
size_t item_size2;
uint8_t *item2;
if (in_isr) {
item2 = (uint8_t *)xRingbufferReceiveUpToFromISR(handle, &item_size2, expected_size - item_size);
} else {
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
item2 = (uint8_t *)xRingbufferReceiveUpTo(handle, &item_size2, ticks_to_wait, expected_size - item_size);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//= (uint8_t *)xRingbufferReceiveUpTo(handle, &item_size2, ticks_to_wait, expected_size - item_size);
TEST_ASSERT_MESSAGE(item2 != NULL, "Failed to receive item");
TEST_ASSERT_MESSAGE(item_size + item_size2 == expected_size, "Total item size is incorrect");
for (int i = 0; i < item_size2; i++) {
TEST_ASSERT_MESSAGE(item2[i] == expected_data[item_size + i], "Item data is invalid");
}
if (in_isr) {
vRingbufferReturnItemFromISR(handle, (void *)item2, NULL);
} else {
vRingbufferReturnItem(handle, (void *)item2);
}
} else {
TEST_ASSERT_MESSAGE(item_size == expected_size, "Item size is incorrect");
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
}
/* ----------------- Basic ring buffer behavior tests cases --------------------
* The following test cases will test basic send, receive, and wrap around
* behavior of each type of ring buffer. Each test case will do the following
* 1) Send multiple items (nearly fill the buffer)
* 2) Receive and check the sent items (also prepares the buffer for a wrap around
* 3) Send a final item that causes a wrap around
* 4) Receive and check the wrapped item
*/
TEST_CASE("Test ring buffer No-Split", "[freertos]")
{
//Create buffer
RingbufHandle_t buffer_handle = xRingbufferCreate(BUFFER_SIZE, RINGBUF_TYPE_NOSPLIT);
TEST_ASSERT_MESSAGE(buffer_handle != NULL, "Failed to create ring buffer");
//Calculate number of items to send. Aim to almost fill buffer to setup for wrap around
int no_of_items = (BUFFER_SIZE - (ITEM_HDR_SIZE + SMALL_ITEM_SIZE)) / (ITEM_HDR_SIZE + SMALL_ITEM_SIZE);
//Test sending items
for (int i = 0; i < no_of_items; i++) {
send_item_and_check(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
//Test receiving items
for (int i = 0; i < no_of_items; i++) {
receive_check_and_return_item_no_split(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Write pointer should be near the end, test wrap around
uint32_t write_pos_before, write_pos_after;
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_before, NULL);
//Send large item that causes wrap around
send_item_and_check(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
//Receive wrapped item
receive_check_and_return_item_no_split(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_after, NULL);
TEST_ASSERT_MESSAGE(write_pos_after < write_pos_before, "Failed to wrap around");
//Cleanup
vRingbufferDelete(buffer_handle);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_CASE("Test ring buffer Allow-Split", "[freertos]")
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Create buffer
RingbufHandle_t buffer_handle = xRingbufferCreate(BUFFER_SIZE, RINGBUF_TYPE_ALLOWSPLIT);
TEST_ASSERT_MESSAGE(buffer_handle != NULL, "Failed to create ring buffer");
//Calculate number of items to send. Aim to almost fill buffer to setup for wrap around
int no_of_items = (BUFFER_SIZE - (ITEM_HDR_SIZE + SMALL_ITEM_SIZE)) / (ITEM_HDR_SIZE + SMALL_ITEM_SIZE);
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Test sending items
for (int i = 0; i < no_of_items; i++) {
send_item_and_check(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
//Test receiving items
for (int i = 0; i < no_of_items; i++) {
receive_check_and_return_item_allow_split(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
//Write pointer should be near the end, test wrap around
uint32_t write_pos_before, write_pos_after;
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_before, NULL);
//Send large item that causes wrap around
send_item_and_check(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
//Receive wrapped item
receive_check_and_return_item_allow_split(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_after, NULL);
TEST_ASSERT_MESSAGE(write_pos_after < write_pos_before, "Failed to wrap around");
//Cleanup
vRingbufferDelete(buffer_handle);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_CASE("Test ring buffer Byte Buffer", "[freertos]")
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Create buffer
RingbufHandle_t buffer_handle = xRingbufferCreate(BUFFER_SIZE, RINGBUF_TYPE_BYTEBUF);
TEST_ASSERT_MESSAGE(buffer_handle != NULL, "Failed to create ring buffer");
//Calculate number of items to send. Aim to almost fill buffer to setup for wrap around
int no_of_items = (BUFFER_SIZE - SMALL_ITEM_SIZE) / SMALL_ITEM_SIZE;
//Test sending items
for (int i = 0; i < no_of_items; i++) {
send_item_and_check(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
//Test receiving items
for (int i = 0; i < no_of_items; i++) {
receive_check_and_return_item_byte_buffer(buffer_handle, small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
//Write pointer should be near the end, test wrap around
uint32_t write_pos_before, write_pos_after;
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_before, NULL);
//Send large item that causes wrap around
send_item_and_check(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
//Receive wrapped item
receive_check_and_return_item_byte_buffer(buffer_handle, large_item, LARGE_ITEM_SIZE, TIMEOUT_TICKS, false);
vRingbufferGetInfo(buffer_handle, NULL, NULL, &write_pos_after, NULL);
TEST_ASSERT_MESSAGE(write_pos_after < write_pos_before, "Failed to wrap around");
//Cleanup
vRingbufferDelete(buffer_handle);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
/* ----------------------- Ring buffer queue sets test ------------------------
* The following test case will test receiving from ring buffers that have been
* added to a queue set. The test case will do the following...
* 1) Ring buffer of each type is created and added to the queue set
* 2) A receiving task is created to select from the queue set and read from the appropriate ring buffer
*/
static void queue_set_receiving_task(void *queue_set_handle)
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
QueueSetHandle_t queue_set = (QueueSetHandle_t)queue_set_handle;
//Receive multiple items via queue set
BaseType_t done = pdFALSE;
int no_of_items = BUFFER_SIZE / SMALL_ITEM_SIZE;
int items_rec_count[NO_OF_RB_TYPES] = {0};
while (done != pdTRUE) {
xQueueSetMemberHandle member = xQueueSelectFromSet(queue_set, TIMEOUT_TICKS);
//Read from selected ring buffer
if (xRingbufferCanRead(buffer_handles[0], member) == pdTRUE) {
//No-split buffer
receive_check_and_return_item_no_split(buffer_handles[0], small_item, SMALL_ITEM_SIZE, 0, false);
items_rec_count[0] ++;
} else if (xRingbufferCanRead(buffer_handles[1], member) == pdTRUE) {
//Allow-split buffer
receive_check_and_return_item_allow_split(buffer_handles[1], small_item, SMALL_ITEM_SIZE, 0, false);
items_rec_count[1] ++;
} else if (xRingbufferCanRead(buffer_handles[2], member) == pdTRUE){
//Byte buffer
receive_check_and_return_item_byte_buffer(buffer_handles[2], small_item, SMALL_ITEM_SIZE, 0, false);
items_rec_count[2] ++;
} else {
TEST_ASSERT_MESSAGE( false, "Error with queue set member");
}
//Check for completion
if (items_rec_count[0] == no_of_items &&
items_rec_count[1] == no_of_items &&
items_rec_count[2] == no_of_items) {
done = pdTRUE;
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
xSemaphoreGive(done_sem);
vTaskDelete(NULL);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_CASE("Test ring buffer with queue sets", "[freertos]")
{
QueueSetHandle_t queue_set = xQueueCreateSet(NO_OF_RB_TYPES);
done_sem = xSemaphoreCreateBinary();
//Create ring buffer of each type, then add them to a queue set
for (int i = 0; i < NO_OF_RB_TYPES; i++) {
buffer_handles[i] = xRingbufferCreate(BUFFER_SIZE, i);
TEST_ASSERT_MESSAGE(buffer_handles[i] != NULL, "Failed to create ring buffer");
TEST_ASSERT_MESSAGE(xRingbufferAddToQueueSetRead(buffer_handles[i], queue_set) == pdPASS, "Failed to add to read queue set");
}
//Create a task to send items to each ring buffer
int no_of_items = BUFFER_SIZE / SMALL_ITEM_SIZE;
xTaskCreatePinnedToCore(queue_set_receiving_task, "rec tsk", 2048, (void *)queue_set, UNITY_FREERTOS_PRIORITY + 1 , NULL, 0);
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Send multiple items to each type of ring buffer
for (int i = 0; i < no_of_items; i++) {
for (int j = 0; j < NO_OF_RB_TYPES; j++) {
send_item_and_check(buffer_handles[j], small_item, SMALL_ITEM_SIZE, TIMEOUT_TICKS, false);
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
xSemaphoreTake(done_sem, portMAX_DELAY);
vSemaphoreDelete(done_sem);
//Remove and delete ring buffers from queue sets
for (int i = 0; i < NO_OF_RB_TYPES; i++) {
TEST_ASSERT_MESSAGE(xRingbufferRemoveFromQueueSetRead(buffer_handles[i], queue_set) == pdTRUE, "Failed to remove from read queue set");
vRingbufferDelete(buffer_handles[i]);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
vQueueDelete(queue_set);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
/* -------------------------- Test ring buffer ISR -----------------------------
* The following test case tests ring buffer ISR API. A timer is used to trigger
* the ISR. The test case will do the following
* 1) ISR will be triggered periodically by timer
* 2) The ISR will iterate through all ring buffer types where each iteration
* will send then receive an item to a ring buffer.
*/
#define TIMER_GROUP 0
#define TIMER_NUMBER 0
#define ISR_ITERATIONS ((BUFFER_SIZE / SMALL_ITEM_SIZE) * 2)
intr_handle_t ringbuffer_isr_handle;
static int buf_type;
static int iterations;
static void ringbuffer_isr(void *arg)
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Clear timer interrupt
TIMERG0.int_clr_timers.t0 = 1;
TIMERG0.hw_timer[xPortGetCoreID()].config.alarm_en = 1;
//Test sending to buffer from ISR from ISR
if (buf_type < NO_OF_RB_TYPES) {
send_item_and_check(buffer_handles[buf_type], (void *)small_item, SMALL_ITEM_SIZE, 0, true);
}
//Receive item from ISR
if (buf_type == RINGBUF_TYPE_NOSPLIT) {
//Test receive from ISR for no-split buffer
receive_check_and_return_item_no_split(buffer_handles[buf_type], (void *)small_item, SMALL_ITEM_SIZE, 0, true);
buf_type++;
} else if (buf_type == RINGBUF_TYPE_ALLOWSPLIT) {
//Test send from ISR to allow-split buffer
receive_check_and_return_item_allow_split(buffer_handles[buf_type], (void *)small_item, SMALL_ITEM_SIZE, 0, true);
buf_type++;
} else if (buf_type == RINGBUF_TYPE_BYTEBUF) {
//Test receive from ISR for byte buffer
receive_check_and_return_item_byte_buffer(buffer_handles[buf_type], (void *)small_item, SMALL_ITEM_SIZE, 0, true);
buf_type++;
} else if (buf_type == NO_OF_RB_TYPES) {
//Check if all iterations complete
if (iterations < ISR_ITERATIONS) {
iterations++;
buf_type = 0; //Reset and iterate through each buffer type again
return;
} else {
//Signal complete
BaseType_t task_woken = pdFALSE;
xSemaphoreGiveFromISR(done_sem, &task_woken);
if (task_woken == pdTRUE) {
buf_type++;
portYIELD_FROM_ISR();
}
}
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
static void setup_timer()
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
//Setup timer for ISR
int timer_group = TIMER_GROUP;
int timer_idx = TIMER_NUMBER;
timer_config_t config;
config.alarm_en = 1;
config.auto_reload = 1;
config.counter_dir = TIMER_COUNT_UP;
config.divider = 10000;
config.intr_type = TIMER_INTR_LEVEL;
config.counter_en = TIMER_PAUSE;
timer_init(timer_group, timer_idx, &config); //Configure timer
timer_pause(timer_group, timer_idx); //Stop timer counter
timer_set_counter_value(timer_group, timer_idx, 0x00000000ULL); //Load counter value
timer_set_alarm_value(timer_group, timer_idx, 20); //Set alarm value
timer_enable_intr(timer_group, timer_idx); //Enable timer interrupt
timer_set_auto_reload(timer_group, timer_idx, 1); //Auto Reload
timer_isr_register(timer_group, timer_idx, ringbuffer_isr, NULL, 0, &ringbuffer_isr_handle); //Set ISR handler
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
static void cleanup_timer()
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
timer_disable_intr(TIMER_GROUP, TIMER_NUMBER);
esp_intr_free(ringbuffer_isr_handle);
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_CASE("Test ring buffer ISR", "[freertos]")
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
for (int i = 0; i < NO_OF_RB_TYPES; i++) {
buffer_handles[i] = xRingbufferCreate(BUFFER_SIZE, i);
}
done_sem = xSemaphoreCreateBinary();
buf_type = 0;
iterations = 0;
setup_timer();
//Start timer to trigger ISR
timer_start(TIMER_GROUP, TIMER_NUMBER);
//Wait for ISR to complete multiple iterations
xSemaphoreTake(done_sem, portMAX_DELAY);
//Cleanup
cleanup_timer();
vSemaphoreDelete(done_sem);
for (int i = 0; i < NO_OF_RB_TYPES; i++) {
vRingbufferDelete(buffer_handles[i]);
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
/* ---------------------------- Test ring buffer SMP ---------------------------
* The following test case tests each type of ring buffer in an SMP fashion. A
* sending task and a receiving task is created. The sending task will split
* a continuous piece of data into items of random length and send it to a ring
* buffer. The receiving task will receive and check those items.
* Every permutation of core pinning of the sending and receiving task will be
* tested.
*/
#define SRAND_SEED 3 //Arbitrarily chosen srand() seed
#define SMP_TEST_ITERATIONS 4
static const char continuous_data[] = {"A_very_long_string_that_will_be_split_into_"
"items_of_random_lengths_and_sent_to_the_ring_"
"buffer._The_maximum_random_length_will_also_"
"be_increased_over_multiple_iterations_in_this"
"_test"};
#define CONT_DATA_LEN sizeof(continuous_data)
#define CONT_DATA_TEST_BUFF_LEN (CONT_DATA_LEN/2) //This will guarantee that the buffer must do a wrap around at some point
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
typedef struct {
RingbufHandle_t buffer;
ringbuf_type_t type;
} task_args_t;
static SemaphoreHandle_t tasks_done;
static SemaphoreHandle_t tx_done;
static SemaphoreHandle_t rx_done;
static void send_to_buffer(RingbufHandle_t buffer, size_t max_item_size)
{
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
for (int iter = 0; iter < SMP_TEST_ITERATIONS; iter++) {
size_t bytes_sent = 0; //Number of data bytes sent in this iteration
size_t next_item_size; //Size of next item to send
while (bytes_sent < CONT_DATA_LEN) {
//Get size of next item
next_item_size = rand() % (max_item_size + 1);
if (next_item_size + bytes_sent > CONT_DATA_LEN) {
next_item_size = CONT_DATA_LEN - bytes_sent;
}
//Send item
TEST_ASSERT_MESSAGE(xRingbufferSend(buffer, (void *)&(continuous_data[bytes_sent]), next_item_size, TIMEOUT_TICKS) == pdTRUE, "Failed to send an item");
bytes_sent += next_item_size;
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
xSemaphoreGive(tx_done);
xSemaphoreTake(rx_done, portMAX_DELAY);
}
}
static void read_from_buffer(RingbufHandle_t buffer, ringbuf_type_t buf_type, size_t max_rec_size)
{
for (int iter = 0; iter < SMP_TEST_ITERATIONS; iter++) {
size_t bytes_rec = 0; //Number of data bytes received in this iteration
while (bytes_rec < CONT_DATA_LEN) {
size_t item_size, item_size2; //Possible for allow split buffers to receive two items
char *item_data, *item_data2;
//Select appropriate receive function for type of ring buffer
if (buf_type == RINGBUF_TYPE_NOSPLIT) {
item_data = (char *)xRingbufferReceive(buffer, &item_size, TIMEOUT_TICKS);
} else if (buf_type == RINGBUF_TYPE_ALLOWSPLIT) {
BaseType_t ret = xRingbufferReceiveSplit(buffer, (void **)&item_data, (void **)&item_data2, &item_size, &item_size2, TIMEOUT_TICKS);
TEST_ASSERT_MESSAGE(ret == pdTRUE, "Failed to receive any item");
} else {
item_data = (char *)xRingbufferReceiveUpTo(buffer, &item_size, TIMEOUT_TICKS, max_rec_size);
}
//Check received item and return it
TEST_ASSERT_MESSAGE(item_data != NULL, "Failed to receive an item");
if (buf_type == RINGBUF_TYPE_BYTEBUF) {
TEST_ASSERT_MESSAGE(item_size <= max_rec_size, "Received data exceeds max size");
}
for (int i = 0; i < item_size; i++) {
//Check item_data is valid
TEST_ASSERT_MESSAGE(item_data[i] == continuous_data[bytes_rec + i], "Received data is corrupted");
}
bytes_rec += item_size;
vRingbufferReturnItem(buffer, item_data);
if (buf_type == RINGBUF_TYPE_ALLOWSPLIT && item_data2 != NULL) {
//Check item_data2 is valid
for (int i = 0; i < item_size2; i++) {
TEST_ASSERT_MESSAGE(item_data2[i] == continuous_data[bytes_rec + i], "Received split data is corrupted");
}
bytes_rec += item_size2;
vRingbufferReturnItem(buffer, item_data2);
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
TEST_ASSERT_MESSAGE(bytes_rec == CONT_DATA_LEN, "Total length of received data is incorrect");
xSemaphoreGive(rx_done);
xSemaphoreTake(tx_done, portMAX_DELAY);
}
}
freeRTOS/Re-factor ring buffers This fixes multiple bugs with ring buffers and re-factors the code. The public API has not changed, however the underlying implementation have various private functions have been changed. The following behavioral changes have been made - Size of ring buffers for No-Split/Allow-Split buffers will not be rounded up to the nearest 32-bit aligned size. This was done to simplify the implementation - Item size for No-Split/Allow-Split buffers will also be rounded up to the nearest 32-bit aligned size. The following bugs have been fixed - In copyItemToRingbufAllowSplit(), when copying an item where the aligned size is smaller than the remaining length, the function does not consider the case where the true size of the item is less than 4 bytes. - The copy functions will automatically wrap around the write pointers when the remaining length of the buffer is not large enough to fit a header, but does not consider if wrapping around will cause an overlap with the read pointer. This will make a full buffer be mistaken for an empty buffer closes #1711 - xRingbufferSend() can get stuck in a infinite loop when the size of the free memory is larger than the needed_size, but too small to fit in the ring buffer due to alignment and extra overhead of wrapping around. closes #1846 - Fixed documentation with ring buffer queue set API - Adding and removing from queue set does not consider the case where the read/write semaphores actually hold a value. The following functions have been deprecated - xRingbufferIsNextItemWrapped() due to lack of thread safety - xRingbufferAddToQueueSetWrite() and xRingbufferRemoveFromQueueSetWrite() as adding the queue sets only work under receive operations. The following functions have been added - xRingbufferReceiveSplit() and xRingbufferReceiveSplitFromISR() as a thread safe way to receive from allow-split buffers - vRingbufferGetInfo() Documentation for ring buffers has also been added.
2018-04-18 17:20:34 +00:00
static void send_task(void *args)
{
RingbufHandle_t buffer = ((task_args_t *)args)->buffer;
size_t max_item_len = xRingbufferGetMaxItemSize(buffer);
//Test sending short length items
send_to_buffer(buffer, 1);
//Test sending mid length items
send_to_buffer(buffer, max_item_len/2);
//Test sending long length items
send_to_buffer(buffer, max_item_len);
vTaskDelete(NULL);
}
static void rec_task(void *args)
{
RingbufHandle_t buffer = ((task_args_t *)args)->buffer;
size_t max_rec_len = xRingbufferGetMaxItemSize(buffer);
//Test receiving short length items
read_from_buffer(buffer, ((task_args_t *)args)->type, 1);
//Test receiving mid length items
read_from_buffer(buffer, ((task_args_t *)args)->type, max_rec_len/2);
//Test receiving long length items
read_from_buffer(buffer, ((task_args_t *)args)->type, max_rec_len);
xSemaphoreGive(tasks_done);
vTaskDelete(NULL);
}
TEST_CASE("Test ring buffer SMP", "[freertos]")
{
ets_printf("size of buf %d\n", CONT_DATA_LEN);
tx_done = xSemaphoreCreateBinary(); //Semaphore to indicate send is done for a particular iteration
rx_done = xSemaphoreCreateBinary(); //Semaphore to indicate receive is done for a particular iteration
tasks_done = xSemaphoreCreateBinary(); //Semaphore used to to indicate send and receive tasks completed running
srand(SRAND_SEED); //Seed RNG
//Iterate through buffer types (No split, split, then byte buff)
for (ringbuf_type_t buf_type = 0; buf_type <= RINGBUF_TYPE_BYTEBUF; buf_type++) {
//Create buffer
task_args_t task_args;
task_args.buffer = xRingbufferCreate(CONT_DATA_TEST_BUFF_LEN, buf_type); //Create buffer of selected type
task_args.type = buf_type;
for (int prior_mod = -1; prior_mod < 2; prior_mod++) { //Test different relative priorities
//Test every permutation of core affinity
for (int send_core = 0; send_core < portNUM_PROCESSORS; send_core++) {
for (int rec_core = 0; rec_core < portNUM_PROCESSORS; rec_core ++) {
ets_printf("Type: %d, PM: %d, SC: %d, RC: %d\n", buf_type, prior_mod, send_core, rec_core);
xTaskCreatePinnedToCore(send_task, "send tsk", 2048, (void *)&task_args, 10 + prior_mod, NULL, send_core);
xTaskCreatePinnedToCore(rec_task, "rec tsk", 2048, (void *)&task_args, 10, NULL, rec_core);
xSemaphoreTake(tasks_done, portMAX_DELAY);
vTaskDelay(5); //Allow idle to clean up
}
}
}
//Delete ring buffer
vRingbufferDelete(task_args.buffer);
vTaskDelay(10);
}
//Cleanup
vSemaphoreDelete(tx_done);
vSemaphoreDelete(rx_done);
vSemaphoreDelete(tasks_done);
}
static IRAM_ATTR __attribute__((noinline)) bool iram_ringbuf_test()
{
bool result = true;
spi_flash_guard_get()->start(); // Disables flash cache
RingbufHandle_t handle = xRingbufferCreate(CONT_DATA_TEST_BUFF_LEN, RINGBUF_TYPE_NOSPLIT);
result = result && (handle != NULL);
xRingbufferGetMaxItemSize(handle);
vRingbufferDelete(handle);
spi_flash_guard_get()->end(); // Re-enables flash cache
return result;
}
TEST_CASE("Test ringbuffer functions work with flash cache disabled", "[freertos]")
{
TEST_ASSERT( iram_ringbuf_test() );
}