[EXT] [RESEND PATCH] Issue: Non-HE capable interface advertising HE

[Deepti Panchal]

Hi Jouni,

My patch will override the user config in case the hardware does not support HE configuration.
An error message will also be shown in the debug logs indicating the hardware is not HE supported.

Can you  please elaborate on what you mean by:
"I would have expected the configuration to be rejected instead of being silently ignored"

Please find the modified patch with the "Signed-off" line:

======================================================================================
>From 1c6e7696df5664b86a5ce16631e04801fe5907a6 Mon Sep 17 00:00:00 2001
From: Deepti Panchal <deepti.panchal at nxp.com>
Date: Mon, 18 Jul 2022 15:59:10 +0530
Subject: [PATCH] Issue: Non-HE capable interface advertising HE
 IEs/Capabilities in Beacon/Probe Response/Association Response frames if
 "ieee80211ax=1" configuration is used

Change Description:
    1. The HE capability for an AP iface was read from the driver, but was not used to update the "drv_flags" for the iface
    2. Change is now added to update the HE capability in the "drv_flags" of the interface
    3. If the interface does not support HE capability, the "iface->conf->ieee80211ax" flag is set to 0
    4. Hence now Beacons/Probe response/Association Response will advertise HE capabilities ONLY if the interface is HE capable

Signed-off-by: Deepti Panchal <deepti.panchal at nxp.com>
---
 src/ap/hw_features.c | 14 +++++++++++++-
 1 file changed, 13 insertions(+), 1 deletion(-)

diff --git a/src/ap/hw_features.c b/src/ap/hw_features.c
index ed5ff41d3..87d7b52bd 100644
--- a/src/ap/hw_features.c
+++ b/src/ap/hw_features.c
@@ -679,7 +679,19 @@ static int ieee80211ac_supported_vht_capab(struct hostapd_iface *iface)
 #ifdef CONFIG_IEEE80211AX
 static int ieee80211ax_supported_he_capab(struct hostapd_iface *iface)
 {
-	return 1;
+    struct hostapd_hw_modes *mode = iface->current_mode;
+    struct he_capabilities *he_cap = &(mode->he_capab[IEEE80211_MODE_AP]);
+
+    /* Check iface HE capability, update drv_flags and HE config setting */
+    if (he_cap && he_cap->he_supported) {
+        wpa_printf(MSG_DEBUG, "iface hw he_supported: %d", he_cap->he_supported);
+        iface->drv_flags |= WPA_DRIVER_FLAGS_HE_CAPABILITIES;
+    } else {
+        wpa_printf(MSG_DEBUG, "iface does not support HE capability");
+        iface->conf->ieee80211ax = 0;
+    }
+
+    return 1;
 }
 #endif /* CONFIG_IEEE80211AX */
==================================================================================

Thanks,
Deepti

-----Original Message-----
From: Jouni Malinen <j at w1.fi> 
Sent: Saturday, November 5, 2022 19:57
To: Deepti Panchal <deepti.panchal at nxp.com>
Cc: hostap at lists.infradead.org
Subject: Re: [EXT] [RESEND PATCH] Issue: Non-HE capable interface advertising HE

Caution: EXT Email

On Thu, Sep 29, 2022 at 10:41:53AM +0000, Deepti Panchal wrote:
> This mail describes the purpose of the attached "hostapd" patch.
> The patch avoids issue when there is mismatch between user's 11ax 
> config and DRV capability
>
> Behavior (without the patch):
> ==========================
>     Non-HE capable interface advertising HE IEs/Capabilities in 
> Beacon/Probe Response/Association Response frames if "ieee80211ax=1" 
> configuration is used
>
> Patch Description:
> ================
>     1. The HE capability for an AP iface was read from the driver, but was not used to update the "drv_flags" for the interface
>     2. Change is now added to update the HE capability in the "drv_flags" of the interface
>     3. If the interface does not support HE capability, the "iface->conf->ieee80211ax" flag is set to 0
>     4. Now Beacons/Probe response/Association Response will advertise 
> HE capabilities ONLY if the interface is HE capable

So this would make hostapd ignore what the configuration is trying it to do? That does not sound correct behavior, i.e., I would have expected the configuration to be rejected instead of being silently ignored.

In any case, I cannot really consider applying this patch without the
Signed-off-by: line in the commit message as described in the CONTRIBUTIONS file.

--
Jouni Malinen                                            PGP id EFC895FA