fix(bcm43xx): add BCM43224 detection and blacklist conflicting drivers by andyholst · Pull Request #10910 · omacom/omarchy

andyholst · 2026-09-09T01:19:09Z

Summary

Add BCM43224 (14e4:4353) WiFi chip detection to fix-bcm43xx.sh
Blacklist open-source drivers (b43, brcmsmac, bcma) that conflict with broadcom-wl-dkms

Changes

install/hardware/fix-bcm43xx.sh — Add BCM43224 (14e4:4353) detection and blacklist
test/shell.d/bcm43xx-wifi-fix-test.sh — 10 tests covering detection, blacklist, and non-Broadcom skip

Why the blacklist is needed

When b43/brcmsmac/bcma load before wl, they claim the BCM43224 device and prevent the proprietary wl driver from attaching. This leaves the interface down or unstable. Blacklisting ensures broadcom-wl-dkms works out of the box.

Why this only affects Broadcom hardware

The blacklist is only written when a Broadcom WiFi chip is detected (14e4:43a0, 14e4:4331, or 14e4:4353). Systems with Intel, Qualcomm, or other WiFi chips never trigger this code path.

Testing

All 10 new tests pass
Verified on physical MacBook Air 5,2 with BCM43224 WiFi

Real-World WiFi Verification

Metric	Before (brcmsmac)	After (broadcom-wl-dkms)
Driver	brcmsmac (open-source)	wl (proprietary)
Link speed	144.4 Mbps	1170 Mbit/s
Signal	-41 dBm	100%

7.8x faster link speed with broadcom-wl-dkms. Clean boot, no conflicts.

Add BCM43224 (14e4:4353) WiFi chip detection for MacBook Air 5,2. When b43/brcmsmac/bcma load before wl, they claim the device and prevent the proprietary wl driver from attaching. Blacklist them in /etc/modprobe.d/broadcom-wl.conf so broadcom-wl-dkms works out of the box on fresh installs. Verified on physical MacBook Air 5,2: 7.8x link speed improvement (144 Mbps → 1170 Mbps) with broadcom-wl-dkms vs brcmsmac.

Omarchy runs linux-omarchy, and the base install already supplies its headers before hardware setup, per agents/skills/install-scripts.md. linux-headers is the stock Arch kernel's, so it pulled in headers for a kernel that is not installed. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

greptile-apps · 2026-10-02T10:16:26Z

Confidence Score: 4/5

[Medium risk] Adds Wi-Fi driver detection for one additional Broadcom chip.

The PR does not appear safe to merge until existing affected installations receive the fix.

Findings

Existing installs miss the fix. ▶
Blacklist affects other adapters. ▶
Existing chips lack behavior tests. ▶

Summary

The PR adds BCM43224 detection and tests its installation path. Since the previous review, it removes the script-managed blacklist and leaves blacklisting to broadcom-wl-dkms.

_{Reviews (2) · Last reviewed commit: "Leave blacklisting the in-kernel Broadco..."}

greptile-apps · 2026-10-02T10:16:29Z


-if (echo "$pci_info" | grep -q "14e4:43a0" || echo "$pci_info" | grep -q "14e4:4331"); then
-  echo "BCM4360 / BCM4331 detected"
+if (echo "$pci_info" | grep -q "14e4:43a0" || echo "$pci_info" | grep -q "14e4:4331" || echo "$pci_info" | grep -q "14e4:4353"); then


Existing installs miss the fix. This detection and blacklist run during installation or an explicit hardware-setup rerun, but no migration applies them to existing systems. Updating an affected MacBook Air therefore leaves its driver conflict unresolved unless the user reruns hardware setup. Add an upgrade migration so existing installations receive the fix.

greptile-apps · 2026-10-02T10:16:29Z

+blacklist b43
+blacklist brcmsmac
+blacklist bcma


Blacklist affects other adapters. If a machine has a second Broadcom adapter that needs b43 or brcmsmac and is not supported by wl, these system-wide blacklist entries prevent its driver from loading. That adapter stops working even though the detected Wi-Fi device may work. Account for other installed devices before applying the blacklist globally.

greptile-apps · 2026-10-02T10:16:30Z

+grep -q '14e4:43a0' "$fix_script" || \
+  fail "fix-bcm43xx.sh still detects BCM4360 (14e4:43a0)"
+pass "fix-bcm43xx.sh still detects BCM4360 (14e4:43a0)"
+
+grep -q '14e4:4331' "$fix_script" || \
+  fail "fix-bcm43xx.sh still detects BCM4331 (14e4:4331)"
+pass "fix-bcm43xx.sh still detects BCM4331 (14e4:4331)"


Existing chips lack behavior tests. These checks only confirm that the BCM4360 and BCM4331 IDs appear somewhere in the script. They would still pass if detection stopped handling either chip but its ID remained in a comment. Run both IDs through the stubbed lspci path so the tests catch regressions in the shared detection branch.

The package already ships /usr/lib/modprobe.d/broadcom-wl-dkms.conf blacklisting b43, b43legacy, bcm43xx, bcma, brcm80211, brcmfmac, brcmsmac and ssb, so the extra /etc file added nothing while wl is installed. Unlike the package's file, it outlived the package: removing broadcom-wl-dkms to fall back to brcmsmac or b43 left both blacklisted and the machine without Wi-Fi after reboot. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Co-Authored-By: Codex Medium <noreply@openai.com>

omarchybot · 2026-10-02T10:21:06Z

Thanks for this, @andyholst. I reviewed it and pushed two commits to your branch. The detection line for BCM43224 (14e4:4353) is unchanged; the rest of the script is back to what quattro already does.

What changed on the branch, and why

install/hardware/fix-bcm43xx.sh installs broadcom-wl-dkms alone again, without linux-headers. Omarchy runs linux-omarchy, and the base install already provides linux-omarchy-headers before hardware setup (agents/skills/install-scripts.md). linux-headers is the stock Arch kernel's package, so it installed headers for a kernel that isn't there.
The /etc/modprobe.d/broadcom-wl.conf blacklist is gone. broadcom-wl-dkms already ships /usr/lib/modprobe.d/broadcom-wl-dkms.conf, which blacklists b43, b43legacy, bcm43xx, bcma, brcm80211, brcmfmac, brcmsmac and ssb. I checked that against the package itself on a worker. So while wl is installed, the extra file added nothing. Unlike the package's file, though, it stays behind when the package is removed: someone who uninstalls broadcom-wl-dkms to go back to brcmsmac would keep brcmsmac blacklisted and have no Wi-Fi after a reboot.
The test now checks that broadcom-wl-dkms is the only package installed and that nothing is written to /etc/modprobe.d. I confirmed that each of these assertions fails against the code it replaced.

What ran, and where

All of this ran on a disposable Omarchy VM with a stubbed lspci:

With a BCM43224 present, quattro installs nothing and this branch installs broadcom-wl-dkms.
test/cli passes.
Your test file passes.
test/shell has four failures: ascii, branding-about-animation, omarchy-kernel-migration and screenshot-sanity. Each one fails the same way on quattro without this change.

The VM has no BCM43224, so whether wl actually binds and performs better on a MacBook Air 5,2 hasn't been tested here.

Second opinion

Codex Medium reviewed the PR. Its independence from my review isn't guaranteed. It agreed about the headers and the redundant blacklist, and it was the one that spotted the leftover file breaking a fallback to brcmsmac. On the pushed head it found nothing left to act on.

For the maintainer

The open question isn't about the code. brcmsmac supports BCM43224 in the kernel, and the table in the description shows it connected. So on quattro this machine already has Wi-Fi, and this PR is a choice to switch it to the proprietary wl driver for performance. That decision is the maintainer's. The 1170 Mbit/s figure can't be a link rate for this chip: BCM43224 is 2x2 802.11n, which tops out around 300 Mbit/s. @andyholst, could you say where that number came from? Could you also confirm that 2dce17a01c19b40e3ac1d615a4403bae3d5e1524 still brings up wl on your Air, now that blacklisting is left to the package?

Related open PRs also change install/hardware/fix-bcm43xx.sh, for other chips: #11655 (BCM4352), #12076 (sysfs detection helper) and #13532 (MacBookAir4,1 exception). Whichever lands later will need to keep the other's IDs.

This now waits on the maintainer.

jandrusk · 2026-10-13T16:18:46Z

Diff only adds PCI 14e4:4353 to the existing broadcom-wl-dkms path — blacklist stays with the package (title is a bit strong).

Please confirm on a MacBook Air 5,2 that wl is the right driver here; #7593 / #13532 are the cautionary tale for shipping wl on Airs where b43/brcmsmac was safer. Nearby chip PRs: #15106 (BCM43228), #11311 (BCM4322 b43-firmware).

andyholst requested review from dhh and ryanrhughes as code owners September 9, 2026 01:19

andyholst force-pushed the fix/macbookair5-2-hardware branch from ccac1ff to 5e5e412 Compare September 9, 2026 01:45

andyholst changed the title ~~fix(hardware): add MacBook Air 5,2 support and BCM43224 WiFi detection~~ fix(bcm43xx): add BCM43224 detection and blacklist conflicting drivers Sep 9, 2026

This was referenced Sep 12, 2026

fix-bcm43xx.sh installs broadcom-wl for BCM4331, which hard-freezes a 2011 MacBook Air where in-kernel b43 works #7593

Open

Kernel 7.2 update leaves BCM4360 Macs offline: wl module missing and DKMS fails to build #9386

Open

andyholst closed this Sep 15, 2026

andyholst reopened this Sep 15, 2026

andyholst force-pushed the fix/macbookair5-2-hardware branch from 5e5e412 to ad56c45 Compare September 15, 2026 20:03

andrew-boyd mentioned this pull request Sep 21, 2026

[Intel Mac P10] Consolidate Broadcom calibration and sleep recovery #12688

Draft

This was referenced Sep 27, 2026

Install matching kernel headers before broadcom-wl-dkms #12090

Closed

Detect BCM43xx, T2, Vulkan, and NVIDIA via sysfs helpers #12076

Open

omarchybot added bug Something isn't working compatibility Make Omarchy work better on everything! labels Sep 27, 2026

llstrk mentioned this pull request Sep 29, 2026

Skip broadcom-wl on MacBookAir4,1 with BCM4331 #13532

Open

greptile-apps Bot reviewed Oct 2, 2026

View reviewed changes

omarchybot mentioned this pull request Oct 2, 2026

Support pre-GCN AMD GPUs and BCM4321 wireless #10936

Open

Navigation Menu

Repository navigation

fix(bcm43xx): add BCM43224 detection and blacklist conflicting drivers - #10910

fix(bcm43xx): add BCM43224 detection and blacklist conflicting drivers#10910
andyholst wants to merge 3 commits into
omacom:quattrofrom
andyholst:fix/macbookair5-2-hardware

andyholst commented Sep 9, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot commented Oct 2, 2026 •

edited

Loading

Uh oh!

greptile-apps Bot Oct 2, 2026

Uh oh!

greptile-apps Bot Oct 2, 2026

Uh oh!

Uh oh!

greptile-apps Bot Oct 2, 2026

Uh oh!

omarchybot commented Oct 2, 2026

Uh oh!

jandrusk commented Oct 13, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

andyholst commented Sep 9, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Changes

Why the blacklist is needed

Why this only affects Broadcom hardware

Testing

Real-World WiFi Verification

Uh oh!

greptile-apps Bot commented Oct 2, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Confidence Score: 4/5

Findings

Summary

Uh oh!

greptile-apps Bot Oct 2, 2026

Choose a reason for hiding this comment

Uh oh!

greptile-apps Bot Oct 2, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

greptile-apps Bot Oct 2, 2026

Choose a reason for hiding this comment

Uh oh!

omarchybot commented Oct 2, 2026

Uh oh!

jandrusk commented Oct 13, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

andyholst commented Sep 9, 2026 •

edited

Loading

greptile-apps Bot commented Oct 2, 2026 •

edited

Loading