Optimized build for the D-Link DIR-860L

I am not sure if I should try if the crash will appear without the wireless setting change I memtioned earlier, after these airtime improvements.

Rekeying caused something in airtime.c to cause a stacktrace, if I read the stack trace correctly it should be this line:

  	 (status->encoding == RX_ENC_HE && streams > 8)))

Which has something to do with data rates for which there are fixes in latest mt76. Could not hurt to try since you can always revert back to a previous build which works fine for you. Disabling rekeying is not recommended for security reasons.
Does it happen on rekeying for an Apple device?

Exactly it happened with Apple devices.

And you are using WPA3? And the Apple device is a MacBook?

It's just WPA2 Enterprise with iPhone.

I am pretty sure disabling rekeying posses less security risks with 801.x.

Hmm, that should work just fine. Could you try using 86400 seconds for the rekeying interval?

Disabling rekeying poses less of security risk but nonetheless it is a security risk. In theory an attacker could more easily crack the wireless encryption.

It's a security risk in the sense of once hacker or whoever cracked the wpa encryption, internet trafic in a longer period is cracked, without rekeying.

With 801.x each device can have it's own key, thus hackers cracked one does not mean they get data from other devices cracked.

Edit: I will report back if the crash appears.

The crash is still there:

[24379.963238] ------------[ cut here ]------------
[24379.972683] WARNING: CPU: 1 PID: 1000 at backports-5.10.42-1/net/mac80211/airtime.c:457 ieee80211_ocb_leave+0x564/0x644 [mac80211]
[24379.996118] Modules linked in: pppoe ppp_async iptable_nat xt_state xt_nat xt_conntrack xt_REDIRECT xt_MASQUERADE xt_FLOWOFFLOAD xt_CT pppox ppp_generic nf_nat nf_flow_table_hw nf_flow_table nf_conntrack mt76x2e mt76x2_common mt76x02_lib mt76 mac80211 ipt_REJECT cfg80211 xt_time xt_tcpudp xt_multiport xt_mark xt_mac xt_limit xt_comment xt_TCPMSS xt_LOG slhc nf_reject_ipv4 nf_log_ipv4 nf_defrag_ipv6 nf_defrag_ipv4 iptable_mangle iptable_filter ip_tables crc_ccitt compat ledtrig_usbport nf_log_ipv6 nf_log_common ip6table_mangle ip6table_filter ip6_tables ip6t_REJECT x_tables nf_reject_ipv6 sha256_generic libsha256 seqiv jitterentropy_rng drbg hmac ghash_generic gf128mul gcm ctr cmac ccm leds_gpio xhci_plat_hcd xhci_pci xhci_mtk xhci_hcd gpio_button_hotplug usbcore nls_base usb_common
[24380.134117] CPU: 1 PID: 1000 Comm: mt76-tx phy1 Not tainted 5.4.132 #0
[24380.147096] Stack : 00000000 8007de24 80740000 8073bd04 807a0000 8073bccc 8073ab68 86b2fc94
[24380.163720]         808e0000 87f368e8 80788d63 806c9ca4 00000001 00000001 86b2fc38 b3edbead
[24380.180347]         00000000 00000000 80920000 00000000 00000030 00000138 746e6961 35206465
[24380.196972]         00000000 000004bc 00000000 00020be5 00000000 807a0000 00000000 86a58b60
[24380.213598]         00000009 00000014 00000004 00000002 00000001 8038e0f8 00000004 808e0004
[24380.230219]         ...
[24380.235074] Call Trace:
[24380.239955] [<8000b64c>] show_stack+0x30/0x100
[24380.248811] [<806063e0>] dump_stack+0xa4/0xdc
[24380.257481] [<8002c130>] __warn+0xc0/0x10c
[24380.265639] [<8002c1d8>] warn_slowpath_fmt+0x5c/0xac
[24380.275644] [<86a58b60>] ieee80211_ocb_leave+0x564/0x644 [mac80211]
[24380.288380] ---[ end trace e6a46e2a94c2d942 ]---

Any chance to get some update on this optimized build? The stable version of the 21.02.0 out there. Maybe there are some advantages of the new version.

1 Like

I moved to official 21.02, it is working well in everyday use (working from home during the day, torrenting during the night). I didn't test WPA3.

An optimized build upstremed would be more than welcome :smiley:
Let's wait for @Bartvz 's feedback

2 Likes

@Bartvz, do you have any trouble trying to get a working build from the latest master codes?

I am asking because I've got serious trouble with master snapshot builds, recent builds made the WiFi chip just disappeared. Not sure about the latest latest, since I have to revert to 21.02.1 release for usability.

Works fine for me using 5.4 kernel

Regarding performance I'm seeing around 300mbit in WDS mode (WPA3) with a few simple optmization patches applied to source. That seems to be the limit for the SoC...

Christmas came sort of early, r18381 is here.
My server literally exploded (PSU) and that took a while to get around to. Also my work has kept me extremely busy so no free time to tinker with stuff

! Before flashing !
This has not been tested on real hardware so only flash if you are able to restore the router to a working state.
Also, since there are lots of changes please do not keep your current configuration when upgrading. Start from scratch.

What's new?

  • Upstream changes, most notably Linux kernel 5.10 and the latest mt76 driver
  • Compiled with GCC 11 and binutils 2.37
5 Likes

Thank you Bart!!! Merry Christmas openwrters :smiley: :star2:

So, I had finally time to flash it and so far (12h) it seems stable. What are your experiences?

3 Likes

Hello, I confirm that r18381 is working very well! I reactivated also the 2.4g wireless (I disabled it some time ago) for connecting an old device, it is as well fast and stable.

2 Likes

After flashing r18381, mine says:
OpenWrt SNAPSHOT r16249-9ac47ee469 / LuCI Master git-21.077.59320-d1bf56d

That does not seem like a match?

1 Like

second flash and it appears as it should.

1 Like

Working nice :slight_smile:
Thank you :smiley:

1 Like

Looks like the Possible Kernel 5.10 regression issue with MT7621 and SW/HW offload enabled problem going to be resolved soon.

It already works with DHCP. PPPoE also works, but it needs a manual step to drop down sirq to 0%!

3 Likes