Date: Sun, 21 Sep 2025 10:17:51 -0700 From: Adrian Chadd <adrian@freebsd.org> To: Gleb Smirnoff <glebius@freebsd.org> Cc: bz@freebsd.org, src-committers@freebsd.org, dev-commits-src-all@freebsd.org, dev-commits-src-main@freebsd.org Subject: Re: git: 785edcc2af5a - main - net80211: convert the rest of the native net80211 drivers to SEQNO_OFFLOAD Message-ID: <CAJ-Vmo=t8Zdjz7%2B_-pE8x0tk=XQASpAfX%2BvFGk8M0Ess9s_W9Q@mail.gmail.com> In-Reply-To: <aM-3ToTQy0MloNXA@cell.glebi.us> References: <202509200047.58K0lcgE017026@gitrepo.freebsd.org> <aM-3ToTQy0MloNXA@cell.glebi.us>
next in thread | previous in thread | raw e-mail | index | archive | help
[-- Attachment #1 --] On Sun, 21 Sept 2025 at 01:29, Gleb Smirnoff <glebius@freebsd.org> wrote: > Hi, > > On Sat, Sep 20, 2025 at 12:47:38AM +0000, Adrian Chadd wrote: > A> commit 785edcc2af5ae32c24f5caf5b9552f62bdae33ee > A> Author: Adrian Chadd <adrian@FreeBSD.org> > A> AuthorDate: 2025-06-10 04:22:10 +0000 > A> Commit: Adrian Chadd <adrian@FreeBSD.org> > A> CommitDate: 2025-09-20 00:42:45 +0000 > A> > A> net80211: convert the rest of the native net80211 drivers to > SEQNO_OFFLOAD > A> > A> * Convert the rest of the drivers to implement driver/offloaded > sequence > A> number handling. > A> > A> * For drivers that implement their own sequence number space > handling > A> for A-MPDU, only call ieee80211_output_seqno_assign() if the frame > A> isn't tagged with M_AMPDU_MPDU, which mirrors the original > net80211 > A> sequence number behaviour. (Except of course, the assignment is > A> now happening during final encap/transmit, not early in encap.) > A> > A> Locally tested (sta mode): > A> > A> * ath > A> * iwn > A> * bwi > A> * bwn > A> * iwm > A> * otus > A> * ral > A> > A> Differential Revision: https://reviews.freebsd.org/D50772 > A> Okayed by: bz > > This commit broke my iwm(4) on Thinkpad X1 Carbon 7th Gen. > > The symptoms are that the interface can associate and even receive DHCP > address, but after that all transmitted packets are lost. Any > communication > attempt results in ARP requests that are not seen by other devices on the > network. Reverting this commit resolves the problem. > > I will be able to test patches tomorrow, please send them to me. > > If we are not able to resolve that before start of the stabweek which is > this > Monday, I'd like to get this reverted from main. > Yup, I figured it out pretty quickly after it was first reported. See: https://reviews.freebsd.org/D52652 It looks like different generations of firmware/chipsets have different expectations for sequence numbers being present. I'm going to land this once kevans (and/or you?) test it out and OK it. Sorry! -a > > -- > Gleb Smirnoff > [-- Attachment #2 --] <div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">On Sun, 21 Sept 2025 at 01:29, Gleb Smirnoff <<a href="mailto:glebius@freebsd.org">glebius@freebsd.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"> Hi,<br> <br> On Sat, Sep 20, 2025 at 12:47:38AM +0000, Adrian Chadd wrote:<br> A> commit 785edcc2af5ae32c24f5caf5b9552f62bdae33ee<br> A> Author: Adrian Chadd <adrian@FreeBSD.org><br> A> AuthorDate: 2025-06-10 04:22:10 +0000<br> A> Commit: Adrian Chadd <adrian@FreeBSD.org><br> A> CommitDate: 2025-09-20 00:42:45 +0000<br> A> <br> A> net80211: convert the rest of the native net80211 drivers to SEQNO_OFFLOAD<br> A> <br> A> * Convert the rest of the drivers to implement driver/offloaded sequence<br> A> number handling.<br> A> <br> A> * For drivers that implement their own sequence number space handling<br> A> for A-MPDU, only call ieee80211_output_seqno_assign() if the frame<br> A> isn't tagged with M_AMPDU_MPDU, which mirrors the original net80211<br> A> sequence number behaviour. (Except of course, the assignment is<br> A> now happening during final encap/transmit, not early in encap.)<br> A> <br> A> Locally tested (sta mode):<br> A> <br> A> * ath<br> A> * iwn<br> A> * bwi<br> A> * bwn<br> A> * iwm<br> A> * otus<br> A> * ral<br> A> <br> A> Differential Revision: <a href="https://reviews.freebsd.org/D50772" rel="noreferrer" target="_blank">https://reviews.freebsd.org/D50772</a><br>; A> Okayed by: bz<br> <br> This commit broke my iwm(4) on Thinkpad X1 Carbon 7th Gen.<br> <br> The symptoms are that the interface can associate and even receive DHCP<br> address, but after that all transmitted packets are lost. Any communication<br> attempt results in ARP requests that are not seen by other devices on the<br> network. Reverting this commit resolves the problem.<br> <br> I will be able to test patches tomorrow, please send them to me.<br> <br> If we are not able to resolve that before start of the stabweek which is this<br> Monday, I'd like to get this reverted from main.<br></blockquote><div><br></div><div>Yup, I figured it out pretty quickly after it was first reported.</div><div><br></div><div>See: <a href="https://reviews.freebsd.org/D52652">https://reviews.freebsd.org/D52652</a></div><div><br></div><div>It looks like different generations of firmware/chipsets have different expectations for</div><div>sequence numbers being present. I'm going to land this once kevans (and/or you?)</div><div>test it out and OK it.</div><div><br></div><div>Sorry!</div><div><br></div><div><br></div><div><br></div><div>-a</div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"> <br> -- <br> Gleb Smirnoff<br> </blockquote></div></div>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-Vmo=t8Zdjz7%2B_-pE8x0tk=XQASpAfX%2BvFGk8M0Ess9s_W9Q>