Long read to the server returns MTU size response

I'm trying to use long read on a characteristics (max size is 500 bytes), with MTU size of 180.

The issue is, that on server side, when BLE_GATT_ACCESS_OP_READ_CHR event is called and I call os_mbuf_append() function (as per example), the server only returns MTU size to the client, even though the long read was called.

I got somewhat working results by calling multiple os_mbuf_appendfrom() where I used MTU size offset for each call (to get the all 500 bytes), but this seemed very sketchy to me. Or is this perhaps the only way to do this?

I could not find any info on long reads with nimble in examples and on the internet, so I'm hoping somebody here can help me.

EDIT: added sample code:

Client does this:

rc = ble_gattc_read_long(connection_handle, chr->chr.val_handle, 0, ble_spp_on_read, NULL);

ble_spp_on_read just prints out the received buffer.

From server event BLE_GATT_ACCESS_OP_READ_CHR

uint16_t test_size = 500;
uint8_t rsp[test_size];
// set test buffer
for(int i = 0; i < test_size; i++) {
if(i > 180) rsp[i] = 2;
else rsp[i] = 1;
printf("%d ", rsp[i]);
}
printf("\n");

struct os_mbuf *txom;
txom= ble_hs_mbuf_from_flat(rsp, test_size);

ret = os_mbuf_append(ctxt->om, txom, test_size);
if(ret == BLE_ATT_ERR_INSUFFICIENT_RES || ret != ESP_OK)
ESP_LOGE(tag, "Error responding: %d", ret);

Thanks!

Start by tracing the BLE_GATT_ACCESS_OP_READ_CHR callback and the client’s ble_gattc_read_long call, then inspect how ctxt->om is populated and consumed across requests. Compare the os_mbuf_append and os_mbuf_appendfrom usage shown in the report. Done means a 500-byte characteristic can be read through the long-read procedure with MTU-sized responses as needed.

Written by the indexing model from the issue text.