kernel stuck at initialization of phaser

ccv

Hi,
we are trying to set up a baseband phaser from Creotech on our Kasli-SoC (Zynq7000). After connecting it to one of the ports and to the CLK on the Kasli-SoC, we flashed our device with

$ artiq_coremgmt config write -f boot <..>/boot.bin

The booting worked, the json files where also updated accordingly and we wrote a new entry in our device_db (are ther any restrictions for the channel_base adress?). So far we were not able to initialize the phaser with .init() as the kernel seems to be stuck while running the rtio_input_data function. I guess this means we are still missing some piece to establish a connection to the phaser. The log shows know anomalies only that the kernel is started. Since our knowledge of ARTIQ is limited we are greatful for any help.

Thanks!

rjo

Wrong device db/gateware.
Use the device db template generator.

ccv

We got it to work, thanks!

ccv

This week it stopped working again.

Today we saw an error message ( up until now the kernel would just hang and be unresponsive):

Core Device Traceback:
Traceback (most recent call first):
File "libksupport/src/rtio_csr.rs", line 84, column 10, in (Rust function)
    <unknown>
            ^
File "phaser_tests.py", line 17, in artiq_run_phaser_tests.Phaser_test.run(..., ...) (RA=+0x14c)
    self.phaser0.init()
File "<artiq>\coredevice\phaser.py", line 477, in ... artiq.coredevice.phaser.Phaser.read8<artiq.coredevice.phaser.Phaser>(...) (RA=+0x1c4)
    rtio_output((self.channel_base << 8) | (addr & 0x7f), 0)
File "<artiq>\coredevice\phaser.py", line 271, in ... artiq.coredevice.phaser.Phaser.init<artiq.coredevice.phaser.Phaser>(...) (inlined)
    board_id = self.read8(PHASER_ADDR_BOARD_ID)
artiq.coredevice.exceptions.DMAError(0): RTIO destination unreachable, output, at 3755364384 mu, channel 0x0028:led0

In our device_db led0 should however not be at this channel :

    device_db["phaser0"] = {
"type": "local",
"module": "artiq.coredevice.phaser",
"class": "Phaser",
"arguments": {
    "channel_base": 0x000028,
    "miso_delay": 1, "gw_rev": 1
   }}

     device_db["led0"] = {
"type": "local",
"module": "artiq.coredevice.ttl",
"class": "TTLOut",
"arguments": {"channel": 0x00002d
 }}

We checked wether we are using the right device_db.py multiple times (with print statements and specifying it when running the experiment with artiq_run). When printing the channel_base of the phaser or the channel of the led0 inside the experiment, we also see values corresponding to our current device database (i.e. 40 and 45). Could artiq be using a cached version of the device_db in the background? The channels for the led0 (and other modules) poping up in error messages like the one above do correspond to an old device_db we used previously

rjo

artiq_rtiomap

ccv

rjo The RTIO channel mapping shows the phaser at the channel specified in the device_db.py file. But the phaser still remains unreactive and the experiment hangs.
It seems like the gateware still mixes up the adresses. We also flashed artiq (we have the Kasli-SoC and used artiq_coremgmt config write -f boot boot.bin to reflash) to fix that but nothing changed.

rjo

Phaser LEDs all ok? Fan running?

ccv

rjo Yes, the leds and the fan were on! We got it to work again now by "downgrading" to artiq=version 8.9. Before we were using a beta version. Still not sure why is suddenly stopped working....maybe someone flashed artiq, but it's unsure why it would go back to the old settings and refuse to to be reflashed. Anyway thank you for the help!