Skip to main content
EGold.4
Associate
Aug 20, 2026 4:59 PM UTC
Solved

STM32_Programmer_CLI.exe v2.23.0 seems to run slower than v2.22.0 when called from an executable.

Problem

STM32_Programmer_CLI.exe -c port=SWD sn=<sn> and -l stlink-only each take ~30 seconds (sometimes much more) under v2.23.0 when launched programmatically by another application (CreateProcess, stdio redirected to pipes, no console window) — the pattern any automated test/programming tool uses. CPU stays near-idle during the delay.

Expected behavior

Same as v2.22.0 and same as running the CLI interactively on v2.23.0: under two seconds. Reinstalling v2.22.0, with no other change to our application, fixes it immediately.

What doesn't explain it

Ruled out through direct testing: console presence, output redirection style, handle inheritance, antivirus, elevation, 32/64-bit calling process, working directory, concurrent vs. sequential calls, a full reboot, network reachability (disabled proxy auto-detect, firewall-blocked the exe, disabled WiFi entirely — none changed the timing), and the local Analytics\TrackedData.json cache (already empty).

A Process Monitor trace of a slow call shows two silent ~15s gaps (no registry/file activity logged) right next to the CLI reading HKCU\Software\JavaSoft\Prefs\com\st\library\analyticscore preferences — but explicitly disabling that preference for v2.23.0 didn't change the timing either.

Reproduction

  1. Multiple ST-LINK probes connected via USB hub.
  2. Launch STM32_Programmer_CLI.exe -l stlink-only (or -c port=SWD sn=<sn>) via CreateProcess with STARTF_USESTDHANDLES + CREATE_NO_WINDOW — not by typing it into cmd.exe, which does not reproduce it.
  3. v2.23.0: ~30s per call. v2.22.0, same launch method: fast.

Can provide the full Process Monitor CSV, or a minimal standalone repro program, on request.

Question

Is this a known issue in v2.23.0, and is there a fix or workaround besides staying on v2.22.0?

 

 

Note: I used Claude throughout this investigation and it also helped me draft this post.

TDK
TDKBest answer
Sep 1, 2026 2:01 PM UTC

Likely related to the absurd 30s timeout introduced in v2.23.

https://community.st.com/stm32cubeprogrammer-mcus-30/v2-23-cli-connection-fail-extremly-slow-on-arm-linux-compared-to-windows-167385?tid=167385&fid=30

"If you feel a post has answered your question, please click ""Accept as Solution""."

4 replies

ST Technical Moderator
Sep 1, 2026 1:47 PM UTC

Hello ​@EGold.4 ,
I was not able to reproduce the timeout that you are encountering.
Is it possible to provide the reproduction program that you have mentioned, along with any additional information that may be useful in this investigation?

In order to give better visibility on the answered topics, please click on 'Accept as Solution' on the reply which solved your issue or answered your question.
TDK
TDKBest answer
Sep 1, 2026 2:01 PM UTC
"If you feel a post has answered your question, please click ""Accept as Solution""."
EGold.4
EGold.4Author
Associate
Sep 2, 2026 12:37 PM UTC

Well connected!

Visitor
Oct 16, 2026 10:20 AM UTC

Hello,

I'm seeing the same behavior on my side, so I'm adding my setup in case it helps reproduce it.

Setup:
- 6 x STLINK-V3MINIE connected to the same PC through a USB hub
- ST-LINK FW  : V3J15M7
- OS: Windows 11
- STM32_Programmer_CLI called from a Python script

Observation:
I have three machines running v2.21.0, v2.22.0 and v2.23.0. With v2.21.0 and v2.22.0, all 6 probes are enumerated in a few seconds. With v2.23.0, the same operation is dramatically slower, at least 50 times longer, with no other change to the setup.
 

Measurements:
- `STM32_Programmer_CLI -l stlink-only`, run manually in a terminal: v2.21.0 [~2 s] / v2.22.0 [~2 s] / v2.23.0 [~180 to 200 s]

Anthony