exo/blue-merle
1
0
Fork 0
Code Issues Pull requests Projects Releases 2 Packages Wiki Activity Actions
2 releases 6 tags
  • v3.1.0-ekaii.1 6cedbd22ef

    v3.1.0-ekaii.1 — Phase-1 patches over upstream v3.0
    Some checks failed
    CI / Build Packages (push) Failing after 22s
    Details
    Stable

    admin_ekaii released this 2026-05-25 17:18:21 +00:00 | 3 commits to ekaii since this release

    Personal/Frederic-hardened build of srlabs/blue-merle v3.0 with Phase-1 patches.

    For Mudi (GL-E750) and Mudi V2 (GL-E750V2) on firmware 4.3.x. Not for Mudi 7 (use exo/blue-merle-mudi7 for that).

    Five bug fixes (each a standalone cherry-pickable commit)

    # Commit Closes
    P1 fix(mac): set locally-administered bit on randomised MAC AUDIT bug #1 — random MAC was identifiable as blue-merle output (unicast & LA=0)
    P2 fix(mac): persist glconfig.macclone_addr across reboots AUDIT bug #2 — uci set glconfig.* set but never committed
    P3 fix(generator): use random.choices for IMEI tail entropy AUDIT bug #3 — random.sample capped tail at 151k unique values (re-introduces upstream PR #64, accidentally reverted by #70)
    P4 fix(validate_imei): off-by-one in Luhn AUDIT bug #4 — canonical example 490154203237518 returned False
    P5 fix(set_imei): settle delay + retry + body-level compare AUDIT bug #5 — read-back raced the EFS flush; brittle byte-equality

    Install

    MUDI=root@192.168.8.1
ssh $MUDI 'gl_modem AT AT+GSN' | tee ~/factory-imei.txt   # CRITICAL: keep this
scp -O blue-merle_3.1.0-ekaii.1-240715.80058_mips_24kc.ipk $MUDI:/tmp/
ssh $MUDI 'opkg install /tmp/blue-merle_3.1.0-ekaii.1-240715.80058_mips_24kc.ipk'
ssh $MUDI 'blue-merle'

    Built with the OpenWrt 23.05.4 ath79/generic SDK against upstream tag v3.0.

    SHA-256

    3ce97eff2a97a356268d49165746decd2ec61ca3fba6fb65d322881ed4552954

    Diff vs upstream v3.0

    See git log v3.0..v3.1.0-ekaii.1 in this repo. 5 fix commits + 1 Makefile bump.

    What's NOT in this build

    • Mudi 7 (E5800) support → see forgejo.ekaii.fr/exo/blue-merle-mudi7
    • Factory-IMEI runtime restore CLI → philosophical disagreement with upstream (originals.py lives in the Mudi 7 fork only)
    • BSSID vendor-realistic OUI selection → upstream PR #74 (pasterbele) addresses this and we don't want to overlap
    Downloads