All of lore.kernel.org
 help / color / mirror / Atom feed
From: Baolu Lu <baolu.lu@linux.intel.com>
To: Peter Kiddier <peter@kiddier.org>, linux-wireless@vger.kernel.org
Cc: brcm80211@lists.linux.dev, iommu@lists.linux.dev,
	arend.vanspriel@broadcom.com, regressions@lists.linux.dev
Subject: Re: [REGRESSION] BCM4355 Wi-Fi fails with Intel IOMMU on 7.3.0-2-t2-noble
Date: Wed, 21 Oct 2026 13:57:53 +0800	[thread overview]
Message-ID: <50cac102-9b50-4d0d-af58-8aac51a6d51c@linux.intel.com> (raw)
In-Reply-To: <CA+3E589O46nj=c1Y+Rwk_nPZZKEUjFvMfWaAFmQ2-7De9iUxFw@mail.gmail.com>

On 10/20/26 19:32, Peter Kiddier wrote:
> Hello,
> 
> I am reporting a Wi-Fi regression on a 2018 Apple MacBook Air with a
> T2 chip, running Linux Mint Zena based on Ubuntu 24.04.
> 
> The Broadcom BCM4355 Wi-Fi device works with kernel 7.2.11-3-t2-noble
> but fails with 7.3.0-2-t2-noble. Disabling the Intel IOMMU using
> intel_iommu=off restores Wi-Fi on 7.3.0-2.
> 
> Hardware:
> - Apple MacBook Air 2018 (MacBookAir8,1)
> - Wi-Fi: Broadcom BCM4355 802.11ac Wireless LAN SoC
> - PCI ID: 14e4:43dc (subsystem 106b:0843)
> - Driver: brcmfmac
> - Interface: wlp2s0
> 
> Test results:
> 
> - 7.2.11-3-t2-noble, default kernel command line: Wi-Fi works.
> - 7.3.0-2-t2-noble, default kernel command line: Wi-Fi fails.
> - 7.3.0-2-t2-noble with intel_iommu=off: Wi-Fi works.
> - Reloading brcmfmac on the failing configuration does not restore Wi-Fi.
> 
> On the failing 7.3.0-2 boot, the driver detects the device, loads its
> firmware and creates wlp2s0:
> 
> brcmfmac 0000:02:00.0: enabling device (0000 -> 0002)
> brcmfmac: brcmf_fw_alloc_request: using brcm/brcmfmac4355c1-pcie for
> chip BCM4355/12
> brcmfmac: brcmf_c_process_txcap_blob: TxCap blob found, loading
> brcmfmac: brcmf_c_preinit_dcmds: Firmware: BCM4355/12 wl0: Jul  8 2023
> 17:49:51 version 9.30.503.0.32.5.87 FWID 01-e74b9cb0
> brcmfmac 0000:02:00.0 wlp2s0: renamed from wlan0
> 
> However, Wi-Fi subsequently stops responding. The kernel log reports
> errors including:
> 
> brcmf_msgbuf_query_dcmd: Timeout on response for query command
> brcmf_cfg80211_scan: scan error (-5)
> brcmf_run_escan: error (-5)
> brcmf_msgbuf_tx_ioctl: Failed to reserve space in commonring
> brcmf_cfg80211_get_channel: chanspec failed (-5)
> 
> There is also an error creating the P2P interface:
> 
> brcmf_cfg80211_add_iface: add iface p2p-dev-wlp2s0 type 10 failed: err=-52
> 
> I tried reloading the driver using:
> 
> sudo modprobe -r brcmfmac_wcc
> sudo modprobe -r brcmfmac
> sudo modprobe brcmfmac
> 
> The driver and firmware loaded again, but Wi-Fi remained unusable.
> 
> The firmware files are present under /lib/firmware/brcm/, and the
> firmware version reported by the driver is the same on the working and
> failing kernels.
> 
> The brcmfmac module metadata differs between the kernels:
> 
> Working kernel 7.2.11-3-t2-noble:
> srcversion: 3B6176FDB720DDE8E7C596B
> 
> Failing kernel 7.3.0-2-t2-noble:
> srcversion: 3DCA38D52402541701276CF
> 
> I did not find obvious PCIe AER or fatal errors in the failing boot's
> kernel log.
> 
> It has been suggested testing intel_iommu=off after reloading
> brcmfmac. The driver reload did not help, but booting with
> intel_iommu=off restored Wi-Fi.
> 
> This strongly suggests an interaction between Intel IOMMU handling and
> the BCM4355/brcmfmac driver in the 7.3.0-2-t2-noble kernel.
> 
> I have not tested a vanilla upstream kernel, so I cannot yet confirm
> whether this reproduces outside the T2 kernel packages. Unfortunately,
> I have limited time available for further testing, so I may not be
> able to investigate additional requests promptly. I wanted to report
> the reproducible behaviour and the intel_iommu=off workaround in case
> it helps identify the regression.
> 
> Could you advise on further diagnostics, or whether this should be
> investigated as an IOMMU or brcmfmac regression?
> 
> Thank you.
> 

Can you reproduce this issue with the mainline kernel?

Do you see any IOMMU fault messages in the kernel log? You can check
by running:

   # dmesg | grep DMAR

Have you tried narrowing down the cause with boot options such as
iommu=pt or intel_iommu=on,sm_off?

Thanks,
baolu

      parent reply	other threads:[~2026-10-21  5:57 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-20 11:32 [REGRESSION] BCM4355 Wi-Fi fails with Intel IOMMU on 7.3.0-2-t2-noble Peter Kiddier
2026-10-20 11:49 ` Peter Kiddier
2026-10-20 12:08   ` Aditya Garg
2026-10-21  5:57 ` Baolu Lu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=50cac102-9b50-4d0d-af58-8aac51a6d51c@linux.intel.com \
    --to=baolu.lu@linux.intel.com \
    --cc=arend.vanspriel@broadcom.com \
    --cc=brcm80211@lists.linux.dev \
    --cc=iommu@lists.linux.dev \
    --cc=linux-wireless@vger.kernel.org \
    --cc=peter@kiddier.org \
    --cc=regressions@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.