adc_continuous_read returning ESP_ERR_TIMEOUT (IDFGH-10857)

[puyoulu]

Answers checklist.

• I have read the documentation ESP-IDF Programming Guide and the issue is not addressed there.
• I have updated my IDF branch (master or release) to the latest version and checked that the issue is present there.
• I have searched the issue tracker for a similar issue and not found a similar issue.

IDF version.

v5.0.3

Operating System used.

Linux

How did you build your project?

Command line with idf.py

If you are using Windows, please specify command line type.

None

Development Kit.

esp32-s3-wroom-1

Power Supply used.

External 5V

What is the expected behavior?

adc_continuous_read is quite easy (80% possibility) to return ESP_ERR_TIMEOUT

What is the actual behavior?

adc_continuous_read is quite easy (80% possibility) to always return ESP_ERR_TIMEOUT (conv_frame_size set to 512), after power on reset, and s_adc_dma_intr is then never being called.
This occurs when running v5.0.3, but never occurs on v5.0.2.

I checked the history, seems this bug was introduced by commit 5da5e18 and d8ee45c.
Wish you guys can confirm this.

I tried below patch, and seems can fix:

diff --git a/components/esp_adc/adc_continuous.c b/components/esp_adc/adc_continuous.c
index 643dfef..ed1f760 100644
--- a/components/esp_adc/adc_continuous.c
+++ b/components/esp_adc/adc_continuous.c
@@ -355,6 +355,11 @@ static IRAM_ATTR bool s_adc_dma_intr(adc_continuous_ctx_t *adc_digi_ctx)
         }
     }
 
+    if (status == ADC_HAL_DMA_DESC_NULL) {
+        //start next turns of dma operation
+        adc_hal_digi_start(&adc_digi_ctx->hal, adc_digi_ctx->rx_dma_buf);
+    }
+
     return need_yield;
 }
 
diff --git a/components/hal/adc_hal.c b/components/hal/adc_hal.c
index 3633816..2051ce6 100644
--- a/components/hal/adc_hal.c
+++ b/components/hal/adc_hal.c
@@ -238,7 +238,6 @@ static void adc_hal_digi_dma_link_descriptors(dma_descriptor_t *desc, uint8_t *d
     HAL_ASSERT(((uint32_t)data_buf % 4) == 0);
     HAL_ASSERT((per_eof_size % 4) == 0);
     uint32_t n = 0;
-    dma_descriptor_t *desc_head = desc;
 
     while (eof_num--) {
         uint32_t eof_size = per_eof_size;
@@ -262,7 +261,7 @@ static void adc_hal_digi_dma_link_descriptors(dma_descriptor_t *desc, uint8_t *d
             n++;
         }
     }
-    desc[n-1].next = desc_head;
+    desc[n-1].next = NULL;
 }
 
 void adc_hal_digi_start(adc_hal_dma_ctx_t *hal, uint8_t *data_buf)

Steps to reproduce.

running continuous_read_main.c

Debug Logs.

No response

More Information.

No response

[puyoulu]

I re-checked the code and did some testings, and found out it's because the dma interrupt stops generating during heavy load. I created a PR, it works fine for me, atleat now. Could you help to review it?

[SeanYang-7]

Hello, I have the same problem, what is the cause of this problem?

[Icarus113]

I re-checked the code and did some testings, and found out it's because the dma interrupt stops generating during heavy load. I created a PR, it works fine for me, atleat now. Could you help to review it?

A PR with similar changes have been accepted before. Sorry for the late reply.

Hello, I have the same problem, what is the cause of this problem?

You may upgrade to use a newer version.

[puyoulu]

A PR with similar changes have been accepted before. Sorry for the late reply.

You did not ever read the PR, right? This PR is based on the one you referenced.

Issue can still be reproduced by applying below patch to idf v5.0.4:

--- a/examples/peripherals/adc/continuous_read/main/continuous_read_main.c
+++ b/examples/peripherals/adc/continuous_read/main/continuous_read_main.c
@@ -12,6 +12,7 @@
 #include "freertos/task.h"
 #include "freertos/semphr.h"
 #include "esp_adc/adc_continuous.h"
+#include "nvs_flash.h"
 
 #define EXAMPLE_READ_LEN   256
 #define EXAMPLE_ADC_CONV_MODE           ADC_CONV_SINGLE_UNIT_1
@@ -111,6 +112,14 @@ void app_main(void)
     ESP_ERROR_CHECK(adc_continuous_register_event_callbacks(handle, &cbs, NULL));
     ESP_ERROR_CHECK(adc_continuous_start(handle));
 
+    vTaskDelay(100);
+    ret = nvs_flash_init();
+    if (ret == ESP_ERR_NVS_NO_FREE_PAGES || ret == ESP_ERR_NVS_NEW_VERSION_FOUND) {
+      ESP_ERROR_CHECK(nvs_flash_erase());
+      ret = nvs_flash_init();
+    }
+    nvs_flash_erase_partition("nvs_1");
+
     while(1) {
 
         /**

nvs_1 is something like this in partiton.csv:
nvs_1, data, nvs, 0x1f8000, 32k,

[ZXTube]

If the PR was merged then why do I still have this problem?

I have esp-idf v5.1.2.
For me this problem appeared without me changing anything in my code.
I have a spiffs file and in a while loop I read an adc chunk and write it to the file.
I ran this code plenty of times, only changing code that happens after the while loop.
And one of the times I ran the code and got this error.
After that I consistently get this error every time I run the code.
And the thing that fixed it was either not writing to the file or reformatting the spiffs partition once and then it works every time
The partition was mostly empty when the error happened so I don't think it was caused by it being full.

Any ideas why this is happening.

Thanks in advance