Skip to content

extract-vendor-firmware: husky follow-ups from #722 - #746

Merged
derekross merged 2 commits into
derekross:mainfrom
gitColinX:husky-fw-followups
Nov 15, 2026
Merged

derekross merged 2 commits into
derekross:mainfrom
gitColinX:husky-fw-followups

Conversation

@gitColinX

@gitColinX gitColinX commented Nov 13, 2026 •

Copy link
Copy Markdown
Contributor

@derekross, the small things from your review of #722, in one commit:

  • A readable refusal. BT_NAME gives the Bluetooth check a name where BT_CHIP is a pattern, so husky's refusal now reads "BCM.hcd is not a Pixel 8 Pro (G5SN_V_HK3) Bluetooth patch". The other devices have no BT_NAME and keep their chip name in the message.
  • A wrong cs40l26.wmfw for husky. A new test gives husky a Halo WMFW that isn't the CS40L26's, and it is refused with "cs40l26.wmfw is not a CS40L26 firmware". I checked that it bites, as you did: with husky taken out of the haptics check, this test fails.
  • The device README. The Bluetooth and haptics sentence now comes after the Wi-Fi paragraph it says "also" about. A new line warns that until the 8 Pro has its own address step, Bluetooth comes up on the placeholder address and a pairing made now has to be made again after that step.

Tested

  • tests/extract-vendor-firmware: all pass, including the two husky refusal cases.
  • tools/pr-check origin/main: all ok.

Your question on older factory builds

I've only run the tool on CP3A.261005.005, the build my unit came with. I don't have an older husky image on hand. If it's useful, I can fetch one or two older husky factory images from Google's page, run the tool on each, and post just the ==> lines here.

@derekross

Copy link
Copy Markdown
Owner

Thank you, Colin. All three follow-ups from #722 are in, and they're right: the 8 Pro's Bluetooth refusal now names the phone, every other device keeps its message as it was, and the new haptics case does what it should. I ran the test on main with this merged (all pass) and once with the CS40L26 check turned off in a copy of the tool, and only your new case failed, with its message.

One thing before I merge, and it's because of your other pull request: #752 gives the 8 Pro its address step, and I'm merging it first. After that, the README's last paragraph here ("Until this package has its own Bluetooth address step…") would say the opposite of the package. Could you drop it, or turn it into "pairings made before 0.1.13 have to be made again once", as you offered on #752? While you're in that file, "The kernel side is kernel #156 (a draft)" is out of date: #156 was closed, and the 8 Pro's Wi-Fi landed through kernel #187 on 11-09.

On older factory builds: yes, if you can, run the tool on one or two older husky builds and post just the ==> lines here. That tells us whether the file names have moved between builds, without any of Google's files leaving your machine.

gitColinX and others added 2 commits November 14, 2026 23:25
Derek's review of derekross#722 asked for three small things:

- The refusal named the pattern ("is not a G5SN_V[0-9]*_HK3 Bluetooth
  patch"). BT_NAME gives it a readable name where BT_CHIP is a pattern:
  "BCM.hcd is not a Pixel 8 Pro (G5SN_V<n>_HK3) Bluetooth patch". Other
  devices keep their chip name.
- No test fed husky a wrong cs40l26.wmfw. A Halo WMFW that isn't the
  CS40L26's is now refused for husky; with husky taken out of the
  haptics check, that test fails.
- The device README said "also" before the Wi-Fi paragraph it referred
  to. The sentence follows it now, with a line that pairings made on the
  placeholder Bluetooth address won't survive the address step.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
derekross#752 gave the package its Bluetooth address step, so the paragraph
that said a pairing wouldn't survive "until this package has its own
address step" now says that pairings made before 0.1.13 have to be
made again once. The Wi-Fi line named kernel derekross#156 as a draft; it
merged through kernel derekross#187 on 2026-11-09.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
@gitColinX

gitColinX commented Nov 15, 2026 •

Copy link
Copy Markdown
Contributor Author

@derekross, both done in a new commit, on today's main with #752 in:

The package test and tests/extract-vendor-firmware pass, and tools/pr-check is ok.

On older factory builds: thank you, I'll come back to you on that separately.

@derekross

Copy link
Copy Markdown
Owner

Thank you, Colin, both are in. The pairing paragraph now matches what 0.1.13 does, and the Wi-Fi line points at kernel #187, which is where it landed. I checked that the tool and test changes are the same as the ones I tested last time, ran the test again on today's main with this merged (all pass), and I'm merging it now.

Thanks for the offer on older factory builds. Whenever you get to it, just the ==> lines here are plenty.

@derekross
derekross merged commit 5b58af0 into derekross:main Nov 15, 2026
2 checks passed
@gitColinX

gitColinX commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor Author

@derekross, here's the older-build check you asked for: tools/extract-vendor-firmware -d husky -s G1MNW on main (6c11752) against two older husky factory images from Google's page, each checked against its published SHA-256 first. Both runs passed, each staged the same ten files as CP3A.261005.005, and the Bluetooth check took each build's patch. The ==> lines, without the staging path:

BP2A.250705.008 (Android 16, July 2025):

==> Unpacking husky-bp2a.250705.008/image-husky-bp2a.250705.008.zip from the factory zip
==> Unpacking vendor.img
==> vendor.img is ext4; reading it with debugfs (no mount)
==> Vendor partition: google/husky/husky:16/BP2A.250705.008/13578956:user/release-keys
==> Wi-Fi firmware: Version=24.67106.59.1 (wlan=r1096522 d0 release2)
==> NVRAM: /vendor/firmware/bcmdhd.cal_MMW (SKU G1MNW)
==> Bluetooth patch: G5SN_V17_HK3 REL FW:d665db8b91 CFG:66e1413cd1
==> Haptics firmware: cs40l26.wmfw (59960 bytes) and its wavetable

AP4A.250105.002 (Android 15, January 2025):

==> Unpacking husky-ap4a.250105.002/image-husky-ap4a.250105.002.zip from the factory zip
==> Unpacking vendor.img
==> vendor.img is ext4; reading it with debugfs (no mount)
==> Vendor partition: google/husky/husky:15/AP4A.250105.002/12701944:user/release-keys
==> Wi-Fi firmware: Version=24.67105.49.2 (wlan=r1076922 d0 release2)
==> NVRAM: /vendor/firmware/bcmdhd.cal_MMW (SKU G1MNW)
==> Bluetooth patch: G5SN_V17_HK3 REL FW:c10407L
==> Haptics firmware: cs40l26.wmfw (59480 bytes) and its wavetable

So the names didn't move between Android 15, 16 and CP3A.261005.005: the same BCM.hcd, cs40l26* and four Wi-Fi files are where the map expects them. The module and board string stays G5SN_V17_HK3. The release part after it differs by build, and the January 2025 one is a shorter form (FW:c10407L, no BR:/CFG:), which the check doesn't depend on. The Wi-Fi firmware and cs40l26.wmfw differ in version and size between builds too. I haven't booted the phone with either build's files; this only says the tool stages them.

Google's files stayed on my PC; the staged copies were deleted after each run.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants