public inbox for linux-wireless@vger.kernel.org
 help / color / mirror / Atom feed
* mac80211 HT/VHT MCS check change causing Wi-Fi speed drop (~20 Mbps cap) on iwlwifi
@ 2026-04-24  8:28 Jayden Fowler
  2026-04-24  8:41 ` Johannes Berg
  0 siblings, 1 reply; 2+ messages in thread
From: Jayden Fowler @ 2026-04-24  8:28 UTC (permalink / raw)
  To: linux-wireless

Hello,

I think I’ve run into a regression in mac80211 that affects Wi-Fi speeds 
on Intel iwlwifi (and possibly some Realtek cards too).

After testing a bunch of kernels, I noticed something pretty consistent:

Kernels with this commit Wi-Fi gets stuck around ~20 Mbps max
Kernels without it (or after removing a small part of it) speeds go back 
to normal (hundreds of Mbps as expected)

The commit I bisected it to is:

574faa0e936d12718e2cadad11ce1e184d9e5a32
"wifi: mac80211: add HT and VHT basic set verification"

The issue seems to come from the new HT/VHT MCS verification logic. When 
it runs, it sometimes decides the AP’s advertised rates are 
“unsupported” and then forces a downgrade in connection mode/bandwidth.

That downgrade looks like it’s what causes the big speed drop.

When I remove this from `net/mac80211/mlme.c`, Wi-Fi goes back to normal 
speeds:

if (conn->mode >= IEEE80211_CONN_MODE_HT &&
     !ieee80211_verify_sta_ht_mcs_support(sdata, sband,
                                          elems->ht_operation)) {
     conn->mode = IEEE80211_CONN_MODE_LEGACY;
     conn->bw_limit = IEEE80211_CONN_BW_LIMIT_20;
     link_id_info(sdata, link_id,
                  "required MCSes not supported, disabling HT\n");
}

if (conn->mode >= IEEE80211_CONN_MODE_VHT &&
     !ieee80211_verify_sta_vht_mcs_support(sdata, link_id, sband,
elems->vht_operation)) {
     conn->mode = IEEE80211_CONN_MODE_HT;
     conn->bw_limit = min_t(enum ieee80211_conn_bw_limit,
                            conn->bw_limit,
                            IEEE80211_CONN_BW_LIMIT_40);
     link_id_info(sdata, link_id,
                  "required MCSes not supported, disabling VHT\n");
}

This is my wifi card i'm testing on Network controller: Intel 
Corporation Dual Band Wireless-AC 3168NGW [Stone Peak] (rev 10)

Thanks, Jayden Fowler.



^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: mac80211 HT/VHT MCS check change causing Wi-Fi speed drop (~20 Mbps cap) on iwlwifi
  2026-04-24  8:28 mac80211 HT/VHT MCS check change causing Wi-Fi speed drop (~20 Mbps cap) on iwlwifi Jayden Fowler
@ 2026-04-24  8:41 ` Johannes Berg
  0 siblings, 0 replies; 2+ messages in thread
From: Johannes Berg @ 2026-04-24  8:41 UTC (permalink / raw)
  To: Jayden Fowler, linux-wireless

On Fri, 2026-04-24 at 04:28 -0400, Jayden Fowler wrote:
> Hello,
> 
> I think I’ve run into a regression in mac80211 that affects Wi-Fi speeds 
> on Intel iwlwifi (and possibly some Realtek cards too).
> 
> After testing a bunch of kernels, I noticed something pretty consistent:
> 
> Kernels with this commit Wi-Fi gets stuck around ~20 Mbps max
> Kernels without it (or after removing a small part of it) speeds go back 
> to normal (hundreds of Mbps as expected)
> 
> The commit I bisected it to is:
> 
> 574faa0e936d12718e2cadad11ce1e184d9e5a32
> "wifi: mac80211: add HT and VHT basic set verification"
> 
> The issue seems to come from the new HT/VHT MCS verification logic. When 
> it runs, it sometimes decides the AP’s advertised rates are 
> “unsupported” and then forces a downgrade in connection mode/bandwidth.
> 
> That downgrade looks like it’s what causes the big speed drop.

Yeah, there are too many AP bugs ...

https://patchwork.kernel.org/project/linux-wireless/patch/99Mv9QEceyPrQhSP52MtAVmz0_kWJmzqotJjD9YW6LGLqk-AZloAueUyHCURilFkuqOh6Ecv8i2KKdSE1ujP3AnbU5QEouVisT1w_V3xdfc=@r26.me/

johannes

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-04-24  8:41 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-04-24  8:28 mac80211 HT/VHT MCS check change causing Wi-Fi speed drop (~20 Mbps cap) on iwlwifi Jayden Fowler
2026-04-24  8:41 ` Johannes Berg

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox