-----Original Message-----
From: yocto@... <yocto@...> On
Behalf Of Andrei Gherzan
Sent: Monday, February 15, 2021 7:20 PM
To: yocto@...
Cc: Andrei Gherzan <andrei.gherzan@...>
Subject: [yocto] [meta-zephy][PATCH 13/14] zephyr-flash-pyocd.bbclass:
Implement configurable probe IDs to program

From: Andrei Gherzan <andrei.gherzan@...>

Implement logic to configure what probes to program based on the
PYOCD_FLASH_IDS variable:
1. by default program all attached probes 2. change default behaviour by
listing the probe IDs to flash

CONNECT_TIMEOUT_SECONDS was also renamed to maintain consistency
with the PYOCD_FLASH_IDS variable.

One can query the IDs using `pyocd list`.

The value of PYOCD_FLASH_IDS can also be injected into the datastore using
BB_ENV_EXTRAWHITE.

Signed-off-by: Andrei Gherzan <andrei.gherzan@...>
---
README.txt | 9 +++++
classes/zephyr-flash-pyocd.bbclass | 57 ++++++++++++++++++++----------
2 files changed, 48 insertions(+), 18 deletions(-)

diff --git a/README.txt b/README.txt
index bda872b..ce5338b 100644
--- a/README.txt
+++ b/README.txt
@@ -67,6 +67,15 @@ dfu-util and/or pyocd need to be installed in your
system. If you observe permission errors or the flashing process seem to
hang, follow those instructions:
https://github.com/pyocd/pyOCD/tree/master/udev

+By default, pyocd tries to flash all the attached probes. This
+behaviour can be customised by defining the PYOCD_FLASH_IDS variable as
+a space-separated list of IDs. Once that is set, the tool will only try
+to program these IDs. You can query for the IDs by running `pyocd list`
+on your host while having the probes attached. Besides setting this
+variable through the build's configuration or metadata, you can also inject
its value from command line with something like:
+
+ $ PYOCD_FLASH_IDS='<ID1> <ID2> <ID3>'
+ BB_ENV_EXTRAWHITE="$BB_ENV_EXTRAWHITE PYOCD_FLASH_IDS"
bitbake
+ <TARGET> -c flash_usb
+
Building and Running Zephyr Tests
=================================
Presently only toolchains for ARM, x86, IAMCU and Nios2 are supported.
diff --git a/classes/zephyr-flash-pyocd.bbclass b/classes/zephyr-flash-
pyocd.bbclass
index 4d24e6a..7e1cec5 100644
--- a/classes/zephyr-flash-pyocd.bbclass
+++ b/classes/zephyr-flash-pyocd.bbclass
@@ -1,4 +1,5 @@
-CONNECT_TIMEOUT_SECONDS ?= "30"
+PYOCD_CONNECT_TIMEOUT_SECONDS ?= "30"
+PYOCD_FLASH_IDS ?= "all"

python do_flash_usb() {
try:
@@ -7,26 +8,46 @@ python do_flash_usb() {
except ImportError:
bb.fatal("Flashing with pyocd needs the relevant python package. Make
sure your host provides it or consult your distribution packages for how to
install this prerequisite.")

- timeout = int(d.getVar('CONNECT_TIMEOUT_SECONDS'))
+ try:
+ timeout = int(d.getVar('PYOCD_CONNECT_TIMEOUT_SECONDS'))
+ except ValueError:
+ bb.fatal(f"PYOCD_CONNECT_TIMEOUT_SECONDS was set to an invalid
+ value: {d.getVar('PYOCD_CONNECT_TIMEOUT_SECONDS')}.")
image = f"{d.getVar('DEPLOY_DIR_IMAGE')}/{d.getVar('PN')}.elf"
- bb.plain(f"Attempting to flash {image} to board {d.getVar('BOARD')}")
+ ids = d.getVar('PYOCD_FLASH_IDS')
+
+ # Compute the list of IDs to program
+ if ids == 'all':
+ ids = []
+ for probe in
ConnectHelper.get_all_connected_probes(blocking=False):
+ ids.append(probe.unique_id)
+ if not ids:
+ bb.fatal("No probe detected. Make sure your target is connected.")
+ else:
+ ids = ids.split()
+ if not ids:
+ bb.fatal("No probe requested for programming. Make sure
+ PYOCD_FLASH_IDS is set.")

- # Try to connect to a probe with a timeout
- now = 0
- step = 3
- while True:
- session = ConnectHelper.session_with_chosen_probe(blocking=False,
return_first=True)
- if session:
- break
- if now >= timeout:
- bb.fatal("Timeout while trying to connect to a probe. Make sure the
target device is connected and the udev is configured accordingly. See
<https://github.com/mbedmicro/pyOCD/tree/master/udev> for help.")
- bb.warn("Can't connect to the probe. Retrying in %d seconds..." % step)
- time.sleep(step)
- now += step
+ # Program each ID
+ for id in ids:
+ bb.plain(f"Attempting to flash {os.path.basename(image)} to
+ board {d.getVar('BOARD')} [{id}]")

- with session:
- FileProgrammer(session).program(image)
- session.board.target.reset()
+ # Try to connect to a probe with a timeout
+ now = 0
+ step = 3
+ while True:
+ session =
ConnectHelper.session_with_chosen_probe(blocking=False,
return_first=True, unique_id=id)
+ if session:
+ break
+ if now >= timeout:
+ bb.fatal(f"Timeout while trying to connect to probe ID: {id}. Make
sure the target device is connected and the udev is configured accordingly.
See <https://github.com/mbedmicro/pyOCD/tree/master/udev> for help.")
+ bb.warn(f"Can't connect to the probe ID: {id}. Retrying in {step}
seconds...")
+ time.sleep(step)
+ now += step
+
+ # Program the sepected probe
+ with session:
+ FileProgrammer(session).program(image)
+ session.board.target.reset()
}

addtask do_flash_usb after do_deploy
--
2.30.1

Is there a way to automatically record build time of a recipe without modifying the recipe itself?  I have a recipe that is a monster, it has many git URIs and produces many packages that are coupled. It should be broken up but company deadlines have kept us from taking the time to do this. I was talking to my boss about this. He said we really needs some concrete data about how much this monster recipe is costing us before green-lighting the massive effort to split up the recipe.  I agree with him, so I would like to know how much time is spent repeatedly building this recipe.  I figure I could do this by adding do_fetch_prepend() and writing the time to a file, and then do_install_append() and writing the time to the file. So after a week or so, I would have many start/end data points to discuss.
However, I did not want to modify the recipe itself. I was hoping to augment the recipe via local.conf. Another bbappend in a new layer would work I guess, but I was hoping there was something simpler. Thanks in advance.