mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Baochen Qiang <baochen.qiang@oss.qualcomm.com>
To: Vitor Soares <ivitro@gmail.com>, Jeff Johnson <jjohnson@kernel.org>
Cc: Vitor Soares <vitor.soares@toradex.com>,
	linux-wireless@vger.kernel.org, ath12k@lists.infradead.org,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH ath-next] wifi: ath12k: provide MAC address list for single-radio wiphys
Date: Tue, 8 Sep 2026 15:18:20 +0800	[thread overview]
Message-ID: <54f89935-526c-4582-a8ca-4f645257db53@oss.qualcomm.com> (raw)
In-Reply-To: <20260902082219.3010793-2-ivitro@gmail.com>



On 9/2/2026 4:22 PM, Vitor Soares wrote:
> From: Vitor Soares <vitor.soares@toradex.com>
> 
> ath12k supports creating vdevs with addresses provided by mac80211.
> Userspace can already make concurrent interfaces work by explicitly
> assigning different locally administered addresses.
> 
> However, ath12k does not advertise an address list, so when a second
> interface is created without an explicit address, mac80211 has nothing
> to pick from and falls back to the permanent address. If the first
> interface is already running with that address, bringing up the second
> one fails with -ENOTUNIQ.
> 
> As in ath11k, provide a list of usable addresses so
> ieee80211_assign_perm_addr() can hand each new interface an unused one.
> Keep the first entry as the radio's own MAC address. Generate the
> remaining entries as locally administered variants by changing only the
> upper nibble of the first octet, which naturally limits the pool to 16
> unique addresses.
> 

We (Qualcomm) ever internally proposed a similar solution, but it turns out to break chips
like QCN9274 etc. I have forwarded this patch to internal developers, will come back once
get a conclusion.

> Handle only single-radio wiphys. Wiphys with more than one radio behave
> exactly as before.
> 
> Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.1.c7-00108-QCAHMTSWPL_V1.0_V2.0_SILICONZ_UPSTREAM-3
> 
> Signed-off-by: Vitor Soares <vitor.soares@toradex.com>
> ---
> Note that a device which is not MLO capable registers one wiphy per
> radio, so each radio of a multi-radio device also gets a pool, seeded
> from that radio's own MAC. I only have single-radio WCN7850 hardware,
> so that case is untested.
> ---
>  drivers/net/wireless/ath/ath12k/mac.c | 48 +++++++++++++++++++++++++++
>  1 file changed, 48 insertions(+)
> 
> diff --git a/drivers/net/wireless/ath/ath12k/mac.c b/drivers/net/wireless/ath/ath12k/mac.c
> index 9a775602775d..bee6701f1029 100644
> --- a/drivers/net/wireless/ath/ath12k/mac.c
> +++ b/drivers/net/wireless/ath/ath12k/mac.c
> @@ -14685,6 +14685,50 @@ static void ath12k_mac_cleanup_unregister(struct ath12k *ar)
>  	kfree(ar->mac.sbands[NL80211_BAND_6GHZ].channels);
>  }
>  
> +static void ath12k_mac_cleanup_mac_address_list(struct ath12k_hw *ah)
> +{
> +	kfree(ah->hw->wiphy->addresses);
> +}
> +
> +static void ath12k_mac_setup_mac_address_list(struct ath12k_hw *ah,
> +					      const u8 *mac_addr)
> +{
> +	struct ath12k *ar = ath12k_ah_to_ar(ah, 0);
> +	struct mac_address *addresses;
> +	u16 n_addresses;
> +	int i;
> +
> +	/* The pool is derived from one base address, so it can describe only a
> +	 * single-radio wiphy: with more radios each radio has its own base
> +	 * address, possibly from a different ath12k_base.
> +	 */
> +	if (ah->num_radio != 1)
> +		return;
> +
> +	/* Only the upper nibble of the first octet is varied, so at most 16
> +	 * addresses can be derived from one base.
> +	 */
> +	n_addresses = min_t(u16, TARGET_NUM_VDEVS(ar->ab), 16);
> +	if (n_addresses <= 1)
> +		return;
> +
> +	addresses = kzalloc_objs(*addresses, n_addresses);
> +	if (!addresses)
> +		return;
> +
> +	ether_addr_copy(addresses[0].addr, mac_addr);
> +	for (i = 1; i < n_addresses; i++) {
> +		ether_addr_copy(addresses[i].addr, mac_addr);
> +		/* set Local Administered Address bit */
> +		addresses[i].addr[0] |= 0x2;
> +
> +		addresses[i].addr[0] += i << 4;
> +	}
> +
> +	ah->hw->wiphy->addresses = addresses;
> +	ah->hw->wiphy->n_addresses = n_addresses;
> +}
> +
>  static void ath12k_mac_hw_unregister(struct ath12k_hw *ah)
>  {
>  	struct ieee80211_hw *hw = ah->hw;
> @@ -14704,6 +14748,7 @@ static void ath12k_mac_hw_unregister(struct ath12k_hw *ah)
>  		ath12k_mac_cleanup_unregister(ar);
>  
>  	ath12k_mac_cleanup_iface_combinations(ah);
> +	ath12k_mac_cleanup_mac_address_list(ah);
>  
>  	SET_IEEE80211_DEV(hw, NULL);
>  }
> @@ -14820,6 +14865,7 @@ static int ath12k_mac_hw_register(struct ath12k_hw *ah)
>  	wiphy->available_antennas_tx = antennas_tx;
>  
>  	SET_IEEE80211_PERM_ADDR(hw, mac_addr);
> +	ath12k_mac_setup_mac_address_list(ah, mac_addr);
>  	SET_IEEE80211_DEV(hw, ab->dev);
>  
>  	ret = ath12k_mac_setup_iface_combinations(ah);
> @@ -15041,6 +15087,8 @@ static int ath12k_mac_hw_register(struct ath12k_hw *ah)
>  		ath12k_mac_cleanup_unregister(ar);
>  	}
>  
> +	ath12k_mac_cleanup_mac_address_list(ah);
> +
>  	SET_IEEE80211_DEV(hw, NULL);
>  
>  	return ret;


      reply	other threads:[~2026-09-08  7:18 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-02  8:22 Vitor Soares
2026-09-08  7:18 ` Baochen Qiang [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=54f89935-526c-4582-a8ca-4f645257db53@oss.qualcomm.com \
    --to=baochen.qiang@oss.qualcomm.com \
    --cc=ath12k@lists.infradead.org \
    --cc=ivitro@gmail.com \
    --cc=jjohnson@kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-wireless@vger.kernel.org \
    --cc=vitor.soares@toradex.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®