[Errno 6] No such device or address
alec-yen opened this issue · comments
Hi, we have 2 ZCU216's. Both ZCU216's were installed in the same configuration.
Using the same micro SD card for both, we run the following command.
from qick import *
soc = QickSoc()
On the 1st ZCU216, it runs properly.
resetting clocks: 245.76 491.52
QICK configuration:
Board: ZCU216
Software version: 0.2.161
Firmware timestamp: Thu Feb 17 16:46:52 2022
Global clocks (MHz): tProcessor 430.080, RF reference 245.760
7 signal generator channels:
0: axis_signal_gen_v4 - tProc output 1, envelope memory 65536 samples
DAC tile 2, blk 0, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
1: axis_signal_gen_v4 - tProc output 2, envelope memory 65536 samples
DAC tile 2, blk 1, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
2: axis_signal_gen_v4 - tProc output 3, envelope memory 65536 samples
DAC tile 2, blk 2, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
3: axis_signal_gen_v4 - tProc output 4, envelope memory 65536 samples
DAC tile 2, blk 3, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
4: axis_signal_gen_v4 - tProc output 5, envelope memory 65536 samples
DAC tile 3, blk 0, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
5: axis_signal_gen_v4 - tProc output 6, envelope memory 65536 samples
DAC tile 3, blk 1, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
6: axis_signal_gen_v4 - tProc output 7, envelope memory 65536 samples
DAC tile 3, blk 2, 32-bit DDS, fabric=430.080 MHz, f_dds=6881.280 MHz
2 readout channels:
0: axis_readout_v2 - controlled by PYNQ
ADC tile 2, blk 0, 32-bit DDS, fabric=307.200 MHz, fs=2457.600 MHz
maxlen 16384 (avg) 1024 (decimated)
triggered by output 0, pin 14, feedback to tProc input 0
1: axis_readout_v2 - controlled by PYNQ
ADC tile 2, blk 2, 32-bit DDS, fabric=307.200 MHz, fs=2457.600 MHz
maxlen 16384 (avg) 1024 (decimated)
triggered by output 0, pin 15, feedback to tProc input 1
7 DACs:
DAC tile 2, blk 0 is 0_230, on JHC3
DAC tile 2, blk 1 is 1_230, on JHC4
DAC tile 2, blk 2 is 2_230, on JHC3
DAC tile 2, blk 3 is 3_230, on JHC4
DAC tile 3, blk 0 is 0_231, on JHC3
DAC tile 3, blk 1 is 1_231, on JHC4
DAC tile 3, blk 2 is 2_231, on JHC3
2 ADCs:
ADC tile 2, blk 0 is 0_226, on JHC7
ADC tile 2, blk 2 is 2_226, on JHC7
4 digital output pins:
0: PMOD0_0_LS (output 0, pin 0)
1: PMOD0_1_LS (output 0, pin 1)
2: PMOD0_2_LS (output 0, pin 2)
3: PMOD0_3_LS (output 0, pin 3)
tProc axis_tproc64x32_x8: program memory 8192 words, data memory 4096 words
external start pin: None
On the 2nd ZCU216, however, it gives an [Errno 6] No such device or address error. The fact that we did not see this error on the 1st FPGA (using the same micro SD card) makes it seem like this is an error with the installation of the FPGA or that the FPGA is defective. Any suggestions for debugging this?
resetting clocks: 245.76 491.52
---------------------------------------------------------------------------
OSError Traceback (most recent call last)
<ipython-input-2-cf6eccdc61bb> in <module>
1 # Load bitstream with custom overlay
----> 2 soc = QickSoc()
3 # Since we're running locally on the QICK, we don't need a separate QickConfig object.
4 # If running remotely, you could generate a QickConfig from the QickSoc:
5 # soccfg = QickConfig(soc.get_cfg())
/home/xilinx/jupyter_notebooks/qick/qick_lib/qick/qick.py in __init__(self, bitfile, force_init_clks, ignore_version, no_tproc, clk_output, external_clk, **kwargs)
266 self.ip_dict['usp_rf_data_converter_0']['parameters'])
267
--> 268 self.config_clocks(force_init_clks)
269
270 # RF data converter (for configuring ADCs and DACs, and setting NCOs)
/home/xilinx/jupyter_notebooks/qick/qick_lib/qick/qick.py in config_clocks(self, force_init_clks)
428 self.download()
429 if not self.clocks_locked():
--> 430 self.set_all_clks()
431 self.download()
432 if not self.clocks_locked():
/home/xilinx/jupyter_notebooks/qick/qick_lib/qick/qick.py in set_all_clks(self)
565 # default value is 0x012C22
566 xrfclk.xrfclk._Config['lmk04828'][lmk_freq][55] = 0x012C02
--> 567 xrfclk.set_ref_clks(lmk_freq=lmk_freq, lmx_freq=lmx_freq)
568 elif self['board'] == 'RFSoC4x2':
569 # master clock generator is LMK04828, always outputs 245.76
/usr/local/share/pynq-venv/lib/python3.8/site-packages/xrfclk/xrfclk.py in set_ref_clks(lmk_freq, lmx_freq)
248
249 for lmk in lmk_devices:
--> 250 _set_LMK_clks(lmk_freq, lmk)
251 for lmx in lmx_devices:
252 _set_LMX_clks(lmx_freq, lmx)
/usr/local/share/pynq-venv/lib/python3.8/site-packages/xrfclk/xrfclk.py in _set_LMK_clks(lmk_freq, lmk)
153 raise RuntimeError("Frequency {} MHz is not valid.".format(lmk_freq))
154 else:
--> 155 _write_LMK_regs(_Config[lmk['compatible']][lmk_freq], lmk)
156
157 def _get_spidev_path(dev):
/usr/local/share/pynq-venv/lib/python3.8/site-packages/xrfclk/xrfclk.py in _write_LMK_regs(reg_vals, lmk)
71 data = struct.pack('>I', v)
72 if lmk['num_bytes'] == 3:
---> 73 f.write(data[1:])
74 else:
75 f.write(data)
OSError: [Errno 6] No such device or address
Hi - this is strange, it looks like the CPU is unable to communicate with the clock chips on the CLK104 daughterboard.
Just to make sure:
- you're using the ZCU216 image from sarafs1926/ZCU216-PYNQ#1?
- you're not doing anything before trying to load the QickSoc? we did once see a similar error, where writing values to the fan controller somehow interfered with access to the clock chips - everything was fine if we rebooted and tried again without messing with the fan controller.
- all the jumpers, switches, etc. on the boards are configured identically?
Here are some things to try (probably in this order), to debug whether this might be a hardware problem:
- Check the connection of the CLK104 board to the ZCU216 - it seems possible that you'd get this problem if the connection was loose, or one of the pins on the high-density connector is bent/damaged.
- If the CLK104 board looks fine, you could swap your two CLK104 boards.
- Xilinx provides a special demo tool which exercises the clocks, DACs, and ADCs (https://xilinx-wiki.atlassian.net/wiki/spaces/A/pages/246153525/RF+DC+Evaluation+Tool+for+ZCU216+board+-+Quick+start), as well as a more general self-test (look for XTP580 on https://www.xilinx.com/products/boards-and-kits/zcu216.html) - you can try running that on your two boards. If it fails on the bad board, you can use that as a starting point when dealing with Xilinx support, maybe you can get a warranty replacement.
Hi, thanks so much for your help! Pinpointing the CLK104 helped a lot with debugging. You guessed correctly, it seems the CLK104 board was slightly loose. Making sure the connection is solid solved the problem - thanks again!
Amazing, glad to hear it!