mirror of
https://github.com/espressif/esp-idf.git
synced 2026-10-01 02:30:52 +03:00
Before: The cache won't be disabled when XIP on psram. But during flash erasing/programming, read data will be courrupt. When XIP in psram is enabled, the image is not mapped to the cache so usually there will be no flash access. The only way to read from flash is via the driver or use mmap. The driver has protection during erasing, while th mmap region not. Now: Mmap APIs provide a flag to make mmap->unmap region mutually exclusive to flash erase/programming when XIP from psram. SPI Flash write APIs will benefit from this. When the flag is used, no concurrent access to mapped region will happen while writing; otherwise the cache will be disable to avoid data corruption. Most ESP-IDF APIs calls mmap with this flag. As for users calling mmap-like APIs directly, they can choose whether to enable this by a flag. Closes https://github.com/espressif/esp-idf/issues/14897
112 lines
3.7 KiB
C
112 lines
3.7 KiB
C
/*
|
|
* SPDX-FileCopyrightText: 2016-2022 Espressif Systems (Shanghai) CO LTD
|
|
*
|
|
* SPDX-License-Identifier: Apache-2.0
|
|
*/
|
|
#include <stdio.h>
|
|
#include <stdlib.h>
|
|
#include "unity.h"
|
|
#include "test_utils.h"
|
|
#include "esp_partition.h"
|
|
|
|
|
|
TEST_CASE("Can read partition table", "[partition]")
|
|
{
|
|
|
|
const esp_partition_t *p = esp_partition_find_first(ESP_PARTITION_TYPE_APP, ESP_PARTITION_SUBTYPE_ANY, NULL);
|
|
TEST_ASSERT_NOT_NULL(p);
|
|
TEST_ASSERT_EQUAL(0x20000, p->address);
|
|
TEST_ASSERT_EQUAL(ESP_PARTITION_SUBTYPE_APP_FACTORY, p->subtype);
|
|
|
|
esp_partition_iterator_t it = esp_partition_find(ESP_PARTITION_TYPE_DATA, ESP_PARTITION_SUBTYPE_ANY, NULL);
|
|
TEST_ASSERT_NOT_NULL(it);
|
|
int count = 0;
|
|
const esp_partition_t* prev = NULL;
|
|
for (; it != NULL; it = esp_partition_next(it)) {
|
|
const esp_partition_t *p = esp_partition_get(it);
|
|
TEST_ASSERT_NOT_NULL(p);
|
|
if (prev) {
|
|
TEST_ASSERT_TRUE_MESSAGE(prev->address < p->address, "incorrect partition order");
|
|
}
|
|
prev = p;
|
|
++count;
|
|
}
|
|
esp_partition_iterator_release(it);
|
|
TEST_ASSERT_EQUAL(5, count);
|
|
|
|
it = esp_partition_find(ESP_PARTITION_TYPE_ANY, ESP_PARTITION_SUBTYPE_ANY, NULL);
|
|
TEST_ASSERT_NOT_NULL(it);
|
|
count = 0;
|
|
for (; it != NULL; it = esp_partition_next(it)) {
|
|
++count;
|
|
}
|
|
esp_partition_iterator_release(it);
|
|
TEST_ASSERT_EQUAL(8, count);
|
|
}
|
|
|
|
TEST_CASE("Can write, read, mmap partition", "[partition][ignore]")
|
|
{
|
|
const esp_partition_t *p = get_test_data_partition();
|
|
printf("Using partition %s at 0x%x, size 0x%x\n", p->label, p->address, p->size);
|
|
TEST_ASSERT_NOT_NULL(p);
|
|
const size_t max_size = 2 * p->erase_size;
|
|
uint8_t *data = (uint8_t *) malloc(max_size);
|
|
TEST_ASSERT_NOT_NULL(data);
|
|
|
|
TEST_ASSERT_EQUAL(ESP_OK, esp_partition_erase_range(p, 0, p->size));
|
|
|
|
srand(0);
|
|
size_t block_size;
|
|
for (size_t offset = 0; offset < p->size; offset += block_size) {
|
|
block_size = ((rand() + 4) % max_size) & (~0x3);
|
|
size_t left = p->size - offset;
|
|
if (block_size > left) {
|
|
block_size = left;
|
|
}
|
|
for (size_t i = 0; i < block_size / 4; ++i) {
|
|
((uint32_t *) (data))[i] = rand();
|
|
}
|
|
TEST_ASSERT_EQUAL(ESP_OK, esp_partition_write(p, offset, data, block_size));
|
|
}
|
|
|
|
srand(0);
|
|
for (size_t offset = 0; offset < p->size; offset += block_size) {
|
|
block_size = ((rand() + 4) % max_size) & (~0x3);
|
|
size_t left = p->size - offset;
|
|
if (block_size > left) {
|
|
block_size = left;
|
|
}
|
|
TEST_ASSERT_EQUAL(ESP_OK, esp_partition_read(p, offset, data, block_size));
|
|
for (size_t i = 0; i < block_size / 4; ++i) {
|
|
TEST_ASSERT_EQUAL(rand(), ((uint32_t *) data)[i]);
|
|
}
|
|
}
|
|
|
|
free(data);
|
|
|
|
const uint32_t *mmap_data;
|
|
esp_partition_mmap_handle_t mmap_handle;
|
|
size_t begin = 3000;
|
|
size_t size = 64000; //chosen so size is smaller than 64K but the mmap straddles 2 MMU blocks
|
|
TEST_ASSERT_EQUAL(ESP_OK, esp_partition_mmap(p, begin, size, ESP_PARTITION_MMAP_DATA | ESP_PARTITION_MMAP_BLOCKS_WRITE,
|
|
(const void **)&mmap_data, &mmap_handle));
|
|
srand(0);
|
|
for (size_t offset = 0; offset < p->size; offset += block_size) {
|
|
block_size = ((rand() + 4) % max_size) & (~0x3);
|
|
size_t left = p->size - offset;
|
|
if (block_size > left) {
|
|
block_size = left;
|
|
}
|
|
for (size_t i = 0; i < block_size / 4; ++i) {
|
|
size_t pos = offset + i * 4;
|
|
uint32_t expected = rand();
|
|
if (pos < begin || pos >= (begin + size)) {
|
|
continue;
|
|
}
|
|
TEST_ASSERT_EQUAL(expected, mmap_data[(pos - begin) / 4]);
|
|
}
|
|
}
|
|
|
|
esp_partition_munmap(mmap_handle);
|
|
}
|