Date: Mon, 20 Mar 2023 20:36:35 +0100 From: Attila Nagy <nagy.attila@gmail.com> To: Eugene Grosbein <eugen@grosbein.net> Cc: freebsd-stable@freebsd.org Subject: Re: Fwd: Kernel DHCP unpredictable/fails (PXE boot), userspace DHCP works just fine Message-ID: <CAM2hQG9eNPKN-=Fv-ALfDvLpFSY7bc8j0M5f%2ByDm9L96LdJb3A@mail.gmail.com> In-Reply-To: <1634f44f-cb69-ee74-5300-4a8c414850f4@grosbein.net> References: <CAM2hQG-p=bfSh_nxuah9zcTBbz7HQ9pYyvOR2f6rC=CUGePKsg@mail.gmail.com> <CAM2hQG-oDRsoccg3S1LykyUF=joWbdJz=GSPOnUroDRxjZ2_iQ@mail.gmail.com> <ab14beaf-f41b-8f45-05f6-fad8a764eba3@grosbein.net> <CAM2hQG-BtvEz649QsFou_sG-QamEdZscCUygKxSuK_C3_MhNow@mail.gmail.com> <1634f44f-cb69-ee74-5300-4a8c414850f4@grosbein.net>
index | next in thread | previous in thread | raw e-mail
[-- Attachment #1 --] Eugene Grosbein <eugen@grosbein.net> ezt írta (időpont: 2023. márc. 19., V, 0:16): > 19.03.2023 5:01, Attila Nagy wrote: > > > Sometimes UEFI/BIOS SETUP has some settings for ACPI/HPET timers > (enable/disable), > > did you try "playing" with such options? > > > > Nope, I haven't thought about that. > > It's enabled (default setting). > > Another possible reason: DHCP packets sent from DHCP client have 1-bit > flag indicating > if client wants to receive an answer as unicast or broadcast UDP packet. > > You could try hacking source to force broadcasts and see if that changes > things. > This NIC's driver has previous history of bad in-chip filtering packets in > transit > from the wire to the host (CPU). > Thanks a lot for the great ideas! Although if I follow the code right, this is already set: https://github.com/freebsd/freebsd-src/blob/c3179891f897d840f578a5139839fcacb587c96d/sys/nfs/bootp_subr.c#L1246 If you have the time to take a look at my patched kernel outputs, I think the response datagrams get to the UDP stack. What I feel is either they doesn't make it to the soreceive call sometimes or the whole process gets reset for some reason. [-- Attachment #2 --] <div dir="ltr"><div class="gmail_quote"><div dir="ltr" class="gmail_attr">Eugene Grosbein <<a href="mailto:eugen@grosbein.net">eugen@grosbein.net</a>> ezt írta (időpont: 2023. márc. 19., V, 0:16):<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">19.03.2023 5:01, Attila Nagy wrote:<br> <br> > Sometimes UEFI/BIOS SETUP has some settings for ACPI/HPET timers (enable/disable),<br> > did you try "playing" with such options?<br> > <br> > Nope, I haven't thought about that.<br> > It's enabled (default setting).<br> <br> Another possible reason: DHCP packets sent from DHCP client have 1-bit flag indicating<br> if client wants to receive an answer as unicast or broadcast UDP packet.<br> <br> You could try hacking source to force broadcasts and see if that changes things.<br> This NIC's driver has previous history of bad in-chip filtering packets in transit<br> from the wire to the host (CPU).<br></blockquote>Thanks a lot for the great ideas!</div><div class="gmail_quote">Although if I follow the code right, this is already set:</div><div class="gmail_quote"><a href="https://github.com/freebsd/freebsd-src/blob/c3179891f897d840f578a5139839fcacb587c96d/sys/nfs/bootp_subr.c#L1246">https://github.com/freebsd/freebsd-src/blob/c3179891f897d840f578a5139839fcacb587c96d/sys/nfs/bootp_subr.c#L1246</a></div><div class="gmail_quote"><br></div><div class="gmail_quote">If you have the time to take a look at my patched kernel outputs, I think the response datagrams get to the UDP stack. What I feel is either they doesn't make it to the soreceive call sometimes or the whole process gets reset for some reason.<br></div><div class="gmail_quote"><br></div></div>home | help
Want to link to this message? Use this
URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAM2hQG9eNPKN-=Fv-ALfDvLpFSY7bc8j0M5f%2ByDm9L96LdJb3A>