Revert "boards/esp32s3-xiao: switch defconfigs to nxinit entrypoint"

[JianyuWang0623]

Summary

Revert commit 7cc6707a76b10cc951f209c09b7f2b6cd228e535 ("boards/esp32s3-xiao: switch defconfigs to nxinit entrypoint").

Felipe Moura de Oliveira reported on the dev mailing list that this commit hangs boot on real Seeed XIAO ESP32-S3 Sense hardware (chip rev v0.2): both defconfigs (combo, usbnsh) never reach nx_start after a normal reset — no console output, no USB enumeration; the board only responds when forced into ROM download mode via the BOOT strap pin. JTAG (OpenOCD + GDB) shows the core statically stuck inside the chip boot ROM address range (0x40034000-0x40048000) across independent halts, while nx_start in that build lives at a completely different address (0x42011068). Reverting only this commit's two defconfig changes on top of current master fixes the boot.

Reference: https://www.mail-archive.com/dev@nuttx.apache.org/msg15063.html

I don't have this specific board (Seeed XIAO ESP32-S3 Sense) locally to reproduce. I did a static before/after build-product comparison (esptool image_info + readelf -S) to sanity-check the most likely hypothesis (image size/segment layout), and it does not explain the hang: both binaries have 3 segments, byte-identical IRAM segment address/size, and the total flash-mapped code/data growth from the reverted commit is ~14.7KB against an 8MB flash budget — negligible. So the real root cause is still open; reverting now to unblock master per the reporter's request, root cause to be investigated further with real hardware access (the reporter offered to help test).

Impact

• esp32s3-xiao combo/usbnsh defconfigs go back to nsh_main as the top-level init entrypoint (pre-nxinit behavior), matching every other released esp32s3-xiao defconfig prior to 7cc6707a76b.
• No other board is affected; the reverted commit only touched these two esp32s3-xiao defconfigs.

Testing

Build-only (no esp32s3-xiao hardware available locally):

$ ./tools/configure.sh -l esp32s3-xiao:usbnsh && make -j8
...
Memory region         Used Size  Region Size  %age Used
             ROM:      210160 B    8388576 B      2.51%
     iram0_0_seg:       34048 B       304 KB     10.94%
     irom0_0_seg:      170607 B    8388576 B      2.03%
     dram0_0_seg:       32268 B       288 KB     10.94%
     drom0_0_seg:      275660 B    8388576 B      3.29%
MKIMAGE: ESP32-S3 binary
Successfully created esp32s3 image.
Generated: nuttx.bin

$ ./tools/configure.sh -l esp32s3-xiao:combo && make -j8
...
MKIMAGE: ESP32-S3 binary
Successfully created esp32s3 image.
Generated: nuttx.bin

$ tools/checkpatch.sh -g HEAD
✔️ All checks pass.

Both defconfigs build cleanly after the revert, matching their pre-7cc6707a76b state (CONFIG_INIT_ENTRYPOINT="nsh_main", no CONFIG_SYSTEM_NXINIT). This is a pure config revert (git revert, no manual edits) so behavior is byte-for-byte the same as before the reverted commit. Runtime boot on real esp32s3-xiao hardware is not re-verified here (no such board available); the reporter (Felipe) has the hardware and offered to help validate.

[github-actions]

MemBrowse Memory Report

No memory changes detected for:

• arduino-mega2560
• esp32-devkitc
• hifive1-revb
• mirtoo
• qemu-armv8a
• qemu-intel64
• rx65n-rsk2mb
• s698pm-dkit
• stm32-nucleo-f103rb