Research window: Past 24 hours (2026-09-05 11:18 ~ 2026-09-06 11:18 Beijing Time) Sources: RT-Thread official news API, GitHub (org: RT-Thread / main repo commits+PR / packages index / topic:rt-thread ecosystem repos), Google News aggregation (zh-CN / en-US, via proxy), Chinese industry media including EET China (see appendix source verification table for details)

Index

• I. Kernel and Version Updates
  • Main repo zero-merge streak extends to third window; new PR #11780 fixes SDIO CIS metadata memory leak (09-05)
• II. Official and Community Activities
  • Official news API sees no new entries for seventeenth consecutive day, elexcon 2026 countdown at 3 days (09-06)
• III. Ecosystem and Vendor Adaptation
  • canopennode-rtt: CiA402 controller-side and diagnostics/EMCY completed, CiA401 generic I/O profile branch launched (09-05 ~ 09-06)
  • Ecosystem reference: gdr GDB debugging framework wraps up FreeRTOS adaptation, RT-Thread retains primary support (09-05, non-dedicated)
• IV. Industry Applications
  • No new entries (no new industry case reports within window)
• V. Summary and Trend Observations
  • Three-day trend (09-04 ~ 09-06)
• Appendix: Source Verification Table

I. Kernel and Version Updates

1.1 Main repo zero-merge streak extends to third window; new PR #11780 fixes SDIO CIS metadata memory leak

Date: 2026-09-05 (PR created 16:59 Beijing Time)　Source: RT-Thread/rt-thread PR #11780 (related issue #11724)

The main rt-thread repo saw zero merges and zero new issues in this window; the most recent merge remains #11776 from 09-03 (third window running), and the release remains v5.2.2 (2025-10-31). The only new addition in the window is PR #11780 “[components][sdio] Fix CIS tuple memory leak” submitted by community developer Huoyanlifusu: it fixes the issue described in #11724 — when sdio_read_cis() processes certain CIS tuples such as CISTPL_FUNCE and CISTPL_VERS_1, it allocates memory but fails to free it after processing, causing memory leaks upon repeated SDIO device init/deinit; the fix releases the temporary tuple after the above two branches complete processing, while unknown tuples continue to be managed by the func->tuples linked list and freed uniformly by sdio_free_cis(). The change touches 1 file with a net addition of 2 lines, and the author verified it on the qemu-vexpress-a9 BSP. This PR is the author’s third consecutive component-level fix (#11756 netdev ping null-pointer crash fix was merged on 08-26; #11774 sal SIOCGIFCONF out-of-bounds fix is still under review), continuing the “component-level memory safety fix” contribution thread. The rest of the review queue is unchanged from yesterday: #11779 CAN framework refactor has no new comments since maintainer CYFS3’s statement on 09-03, and #11775 (c908 rt-smart), #11772 (heap ASan), #11689 (stm32 SPI BDMA) etc. all remain open and untouched.

II. Official and Community Activities

2.1 Official news API sees no new entries for seventeenth consecutive day, elexcon 2026 countdown at 3 days

Date: 2026-09-06　Source: RT-Thread official news API

The official news API (rt_org category) still shows 5 old entries, with the latest remaining the 2026-06-17 Abu Dhabi Investment Forum report, marking the seventeenth consecutive day with no new additions. The next external window is elexcon 2026 in Shenzhen on September 9-11 (RT-Thread exhibiting, “AI hardware full-stack lineup” preview already recorded on 08-30); today is the last weekend before opening, with the countdown at 3 days. Neither Chinese nor English media (Google News via proxy, checked across 1d/2d/3d windows) shows any new RT-Thread coverage within the window; the latest media content remains the 09-04 EET China community project share (recorded yesterday), and the pace of community and media content output has been unaffected during the official channel’s silence.

III. Ecosystem and Vendor Adaptation

3.1 canopennode-rtt: CiA402 controller-side and diagnostics/EMCY completed, CiA401 generic I/O profile branch launched

Dates: 2026-09-05 ~ 09-06 (14:08 ~ 10:06 next day Beijing time, 6 commits)　Source: wdfk-prog/canopennode-rtt (CIA401 branch)

Since yesterday’s 11:08 inclusion of CSP/CSV/CST cyclic synchronous modes, development in this repository has not stopped: the default branch added “CiA402 device diagnostics and EMCY bridge” (09-05 14:08), “Add CiA402 controller support and related tests” (16:13, extending CiA402 coverage from the device side to the controller side), and documentation cleanup/refactoring (16:36); subsequently a new branch CIA401 was opened, and starting 09-05 21:53, “generic I/O device profile core”, “digital event and output fail-safe support”, and the 09-06 10:06 “analog input/output event and fail-safe support” were landed in succession. With this, the CANopenNode port has completed a dual-track expansion within one week since 09-03: motion control profile (CiA402: multi-axis, PP/PV/Homing, CSP/CSV/CST, diagnostics and EMCY, controller side) and generic I/O device profile (CiA401: digital/analog events and fail-safe), with the CAN/industrial fieldbus ecosystem push remaining the most active across the entire community. Combined with the author simultaneously advancing the Generic CAN framework refactor in the main repository (#11779), the “framework layer + application layer” dual-track momentum remains unchanged.

3.2 Ecosystem reference: gdr GDB debugging framework wraps up FreeRTOS adaptation, RT-Thread remains primary support

Date: 2026-09-05 (wrap-up commit at 17:39 Beijing time)　Source: flespark/gdr

The topic:rt-thread repository flespark/gdr (an RTOS debugging assistance framework under the GDB Python environment, providing RTOS command trees such as rtt threads/rtt timers, kernel object pretty-printers, and convenience functions) committed within the window “refactor(freertos): wrapup of new rtos support, code and docs quality enhance”, wrapping up the second FreeRTOS adapter. According to its README: RT-Thread (3.1.x / 4.0.x / 4.1.x, fully validated on Cortex-A9, RV64 since 4.0.4) remains the primary support target; the FreeRTOS channel completed within the window (V10.3.0 ~ V11.3.x) added 64-bit RISC-V lane and MPS2-AN521 dual-core SMP validation on QEMU. This is a general framework extension rather than an RT-Thread-specific commit, but it serves as a positive reference for the GDB debugging tool surface for RT-Thread users (including RISC-V targets), and is included as an ecosystem reference.

IV. Industry Applications

No new entries. There were no new industry application/deployment case reports within the window; the industry side also remained silent over the past three days (09-04 ~ 09-06). The most recently included industry content is the 08-27 Malaysia urban IoT case and the 08-24 RISC-V + AI dual-track summit report (EE Times China). RT-Thread’s latest external developments in embodied intelligence (Lingang Dishui Zhihui) and industrial (Ruiqing industrial platform) directions are expected to be released in a concentrated manner with the opening of elexcon 2026 on 09-09.

V. Summary and Trend Observations

• Kernel side: The main repository’s zero-merge streak enters its third window (most recent merge #11776, 09-03), with only one new PR — #11780, an SDIO CIS tuple memory leak fix (1 file, +2 lines, validated on qemu-vexpress-a9). This is the author’s third consecutive component-level fix, with no other changes in the review queue, release still at v5.2.2, and the accumulation period continuing.
• Ecosystem side: Within one week, canopennode-rtt advanced the CANopenNode port from motion control (full CiA402 modes) to the generic I/O profile (CiA401) and controller-side support, making it the most active development track in the community at present; the GDB debugging assistance framework gdr completed the wrap-up of its second FreeRTOS adapter, with RT-Thread support (including RV64) remaining its primary capability.
• Official/media side: The official news API has been silent for the 17th consecutive day; the media side saw zero new reports in the 24h window (proxy and source checks normal); elexcon 2026 (09-09~11, Shenzhen) is 3 days away, and pre-event silence is an expected rhythm.
• Trend over the past three days (09-04 ~ 09-06): CiA402 multi-axis and PP/PV/Homing, EET reposting a community USB monitoring secondary display project, official silence day 15 (09-04) → env merged WebUI keep_alive (#294), CiA402 cyclic synchronous CSP/CSV/CST landed, official silence day 16 (09-05) → sdio CIS leak fix PR #11780, CiA402 diagnostics/EMCY/controller-side and CiA401 I/O profile branch launched, official silence day 17 (09-06). The current main thrust is “official toolchain WebUI experience iteration + CANopenNode dual-profile ecosystem expansion”, with external activities building up momentum toward elexcon 2026.

Appendix: Source Verification Table