Error codes
When the pod refuses a request, the SDK, the CLI, the MCP server and the apps show a stable code such as BP0001, what went wrong and how to fix it. A code never changes meaning and is never reused, so you can match on it in tests and scripts instead of on the message text.
In Python the code is on the exception: FirmwareError.code (and .values for the values that go with it, a dict). The CLI prints it with every refusal, and benchpod explain BP0001 shows this page's entry offline. Pods with older firmware send no code: see Updating the firmware.
| Code | Message |
|---|---|
| BP0001 | The I/O voltage is not set |
| BP0002 | The pod does not know the command {cmd} |
| BP0003 | An I2C session is already open |
| BP0004 | An SPI session is already open |
| BP0005 | The engine that runs SWD, SPI and I2C is in use, so {want} cannot start |
| BP0006 | The request is missing {arg} |
| BP0007 | The pod has no wiring profile named {name} |
| BP0008 | {name} is not a valid wiring profile name |
| BP0009 | The wiring profile {name} changed since it was loaded (it is now at revision {rev}) |
| BP0010 | The pod cannot hold another wiring profile (at most {max}, and 16 KB in all) |
| BP0011 | The wiring profile is {bytes} bytes; the pod takes at most {max} |
| BP0012 | The wiring profile is not one JSON object |
| BP0013 | {name} is the pod's default wiring profile and cannot be deleted |
| BP0014 | The wiring profile upload skipped ahead (the pod expected offset {expected}) |
| BP0015 | The pod could not read or write its wiring profiles |
| BP0016 | The pod cannot save now: the DAC or a capture is using the bus the storage shares |
| BP0017 | {feature} needs gateware v{min} or later |
| BP0018 | UART rx and tx are both LA{la}; a one-wire UART needs half_duplex |
| BP0019 | half_duplex runs on one pin, but rx is LA{rx} and tx is LA{tx} |
| BP0020 | The UART entry sequence needs the pod's reset pin, which this board does not have |
| BP0021 | The UART entry option {arg} is out of range (0 to {max}) |
| BP0022 | The UART's DE pin LA{la} is also its rx or tx |
| BP0023 | LA{la} has no pull resistor on this board |
| BP0024 | LA{la} supports pull-{have} only on this board |
| BP0025 | {holder} holds this pod ({purpose}, {left_s} s left) |
| BP0026 | The pod's lease was stopped by {by} |
| BP0027 | The pod holds no lease with id {id} |
I/O voltage and pins
The I/O voltage is not set
The pod starts with its I/O voltage unset and never picks one on its own, so it cannot drive a board at a voltage the board does not take. Anything that uses the LA pins (UART, flashing, captures, GPIO, I2C, SPI, sensor emulation) needs it first.
set the board's I/O voltage first: BenchPod(conn, la_voltage=3.3) (1.8 for a 1.8 V board), the benchpod_la_voltage fixture in conftest.py, or bp.set_la_voltage(3.3)
LA{la} has no pull resistor on this board
Only some LA channels have a pull resistor the pod can switch, and which ones depends on the board. la_pins lists each channel's pull and the directions it supports.
Values sent with it: la (number)
pick a channel with a pull resistor (bp.la_pins() shows them), or add one on the board
LA{la} supports pull-{have} only on this board
The pull resistor on LA{la} is wired one way on this board, so it cannot pull {want}. Which way each channel pulls depends on the board; la_pins lists it.
Values sent with it: la (number), want (text), have (text)
use bp.set_pull({la}, "{have}"), or pick a channel that pulls {want} (bp.la_pins() shows them)
Commands and firmware
The pod does not know the command {cmd}
The firmware on the pod is older than the tool that sent the command, or the command name is wrong. Newer SDK, CLI and MCP versions check the pod's capabilities first and say which firmware a feature needs.
Values sent with it: cmd (text)
update the pod's firmware (benchpod flash-self, or the firmware update in the web app)
The request is missing {arg}
A required value was not sent. The SDK, CLI and MCP always send it, so this usually means a hand-written protocol request or a tool older than the firmware.
Values sent with it: arg (text)
pass {arg}; if you did, update the SDK (pip install -U embeddedci)
I2C
An I2C session is already open
The pod runs one I2C master session at a time.
close the open session first (its close(), or leave its with block)
SPI, SWD and the shared engine
An SPI session is already open
The pod runs one SPI master session at a time.
close the open session first (its close(), or leave its with block)
The engine that runs SWD, SPI and I2C is in use, so {want} cannot start
Flashing (SWD), the SPI master and the I2C master share one engine on the pod. While one of them runs, the other two wait.
Values sent with it: want (text)
wait for the running flash, SPI or I2C session to finish (close it), then try again
Wiring profiles
The pod has no wiring profile named {name}
Wiring profiles live on the pod. This one was never saved there, was deleted, or was saved on another pod.
Values sent with it: name (text)
list the pod's profiles with bp.wiring_profiles(), or save this one first with bp.save_wiring(profile, name="{name}")
{name} is not a valid wiring profile name
Profile names are 1 to 40 characters: lowercase letters, digits and dashes, not starting with a dash.
Values sent with it: name (text)
use a name like "nrf52840-dk" or "k66"
The wiring profile {name} changed since it was loaded (it is now at revision {rev})
Someone saved {name} from another connection (the web app, another script or agent) after this one loaded it. Saving now would overwrite their change, so the pod refuses.
Values sent with it: name (text), rev (number)
load it again (bp.load_wiring("{name}")), redo your change and save, or save under another name
The pod cannot hold another wiring profile (at most {max}, and 16 KB in all)
Wiring profiles live on the pod, in a fixed amount of storage.
Values sent with it: max (number)
delete a profile you no longer use (bp.delete_wiring("name")) or replace an existing one
The wiring profile is {bytes} bytes; the pod takes at most {max}
One wiring profile fits in {max} bytes of JSON. Long signal descriptions or many signals add up.
Values sent with it: bytes (number), max (number)
shorten the signal descriptions, or split the board into two profiles
The wiring profile is not one JSON object
The pod stores a profile as the JSON object the tools send. This one is cut off or not an object, which points at a damaged upload or a hand-written request.
save it again; if it keeps failing, check the profile is valid (Wiring.from_dict)
{name} is the pod's default wiring profile and cannot be deleted
A connection that names no profile uses the default, so the pod always keeps one.
Values sent with it: name (text)
make another profile the default first (bp.set_default_wiring("other")), then delete {name}
The wiring profile upload skipped ahead (the pod expected offset {expected})
Profiles are sent to the pod in pieces, in order, starting at offset 0. A piece went missing or another upload started in between.
Values sent with it: expected (number)
save again
The pod could not read or write its wiring profiles
The flash chip that holds the profiles did not answer or a write did not verify. Profiles saved before are kept: a write never replaces them until it has been read back.
try again; if it keeps failing, power-cycle the pod and report it
The pod cannot save now: the DAC or a capture is using the bus the storage shares
The flash chip that holds the wiring profiles shares a bus with the DAC replay and the captures. Saving would stop a running waveform, so the pod refuses instead. Reading profiles always works.
stop the DAC (bp.dac_stop()) or wait for the capture to finish, then save again
firmware
{feature} needs gateware v{min} or later
The pod's FPGA image is older than the feature. Firmware updates bring the matching gateware with them.
Values sent with it: feature (text), min (number)
update the pod's firmware (benchpod flash-self, or the firmware update in the web app)
uart
UART rx and tx are both LA{la}; a one-wire UART needs half_duplex
A normal UART drives its TX pin all the time, so it cannot share the pin with RX. In half duplex the pod only pulls the line low and lets the pull-up hold it high, so both sides can talk on one wire.
Values sent with it: la (number)
pass half_duplex=True for a one-wire UART, or put rx and tx on two LA channels
half_duplex runs on one pin, but rx is LA{rx} and tx is LA{tx}
In half duplex the pod sends and receives on the same wire, so rx and tx name the same LA channel.
Values sent with it: rx (number), tx (number)
give the same LA channel for rx and tx, or leave half_duplex off
The UART entry sequence needs the pod's reset pin, which this board does not have
The entry sequence holds the target in reset with the pod's own reset pin while it holds TX low. Pods before v3 have no reset pin.
use a v3 pod, or open the UART without entry and reset the target yourself
The UART entry option {arg} is out of range (0 to {max})
The entry sequence times are milliseconds, each up to a limit, and the first bytes are at most a few dozen hex bytes.
Values sent with it: arg (text), max (number)
pass {arg} between 0 and {max}
The UART's DE pin LA{la} is also its rx or tx
DE switches an RS-485 transceiver between sending and listening, so it is a wire of its own, from an LA channel to the transceiver's DE and /RE pins.
Values sent with it: la (number)
give de the LA channel wired to the transceiver's DE and /RE, not rx or tx
lease
{holder} holds this pod ({purpose}, {left_s} s left)
Someone took the pod's lease (a CI run, an SDK or CLI session, or a cloud job). While it is held, only its owners may drive the pod; everyone else may still read status and other light reads. The lease ends when its owner releases it, when it is not renewed in time, or when the pod restarts.
Values sent with it: holder (text), purpose (text), left_s (number)
wait until {holder} is done (about {left_s} s); a process of the same run joins its lease with lease_acquire and the same id
The pod's lease was stopped by {by}
This connection or this lease id owned the pod's lease, and {by} stopped it from the web app or the cloud. The command was not run. A connection that owned it hears this once. A command, lease_acquire or lease_renew that carries the stopped lease's id keeps getting it for 30 minutes, so a run that was stopped does not take the pod back; a new id gets a new lease.
Values sent with it: by (text)
the run was stopped on purpose: end it, or take a new lease (lease_acquire) if it should go on
The pod holds no lease with id {id}
lease_renew found no lease: it ran out (it was not renewed before its deadline), it was released, or the pod restarted, which clears it. A lease that was stopped answers BP0026 instead.
Values sent with it: id (text)
take the lease again with lease_acquire; renew it well before its ttl_s runs out