You are not logged in.
Pages: 1
My wireless connection suddenly became very slow(<1Mbps, which should be at least 10 times faster) after an update.
I've tried all the solutions in https://wiki.archlinux.org/index.php/Ne … ss#iwlwifi, but none of them works.
uname -a:
Linux guojize-zhan66-archlinux 5.7.12-arch1-1 #1 SMP PREEMPT Fri, 31 Jul 2020 17:38:22 +0000 x86_64 GNU/Linux
lspci -k:
02:00.0 Network controller: Intel Corporation Wi-Fi 6 AX200 (rev 1a)
Subsystem: Intel Corporation Device 0084
Kernel driver in use: iwlwifi
Kernel modules: iwlwifi
dmesg | grep -F iwlwifi:
[ 2.044038] iwlwifi 0000:02:00.0: enabling device (0000 -> 0002)
[ 2.054947] iwlwifi 0000:02:00.0: api flags index 2 larger than supported by driver
[ 2.054955] iwlwifi 0000:02:00.0: TLV_FW_FSEQ_VERSION: FSEQ Version: 68.3.35.22
[ 2.054960] iwlwifi 0000:02:00.0: Found debug destination: EXTERNAL_DRAM
[ 2.054960] iwlwifi 0000:02:00.0: Found debug configuration: 0
[ 2.055156] iwlwifi 0000:02:00.0: loaded firmware version 53.c31ac674.0 cc-a0-53.ucode op_mode iwlmvm
[ 2.055175] iwlwifi 0000:02:00.0: Direct firmware load for iwl-debug-yoyo.bin failed with error -2
[ 2.190500] iwlwifi 0000:02:00.0: Detected Intel(R) Wi-Fi 6 AX200 160MHz, REV=0x340
[ 2.402853] iwlwifi 0000:02:00.0: base HW address: 74:d8:3e:d1:7a:40
[ 2.418946] Modules linked in: ccm amd64_edac_mod(-) algif_aead edac_mce_amd kvm_amd cbc kvm des_generic libdes ecb arc4 algif_skcipher cmac irqbypass md4 algif_hash crct10dif_pclmul crc32_pclmul af_alg ghash_clmulni_intel aesni_intel iwlmvm crypto_simd cryptd glue_helper amdgpu(+) mac80211 libarc4 snd_hda_codec_realtek pcspkr snd_hda_codec_generic input_leds iwlwifi ledtrig_audio snd_hda_codec_hdmi snd_hda_intel gpu_sched nls_iso8859_1 nls_cp437 i2c_algo_bit snd_intel_dspcfg ttm snd_hda_codec vfat fat drm_kms_helper snd_hda_core cfg80211 cec k10temp snd_hwdep rc_core sp5100_tco snd_pcm i2c_piix4 r8169 syscopyarea ccp sysfillrect snd_timer realtek sysimgblt ucsi_acpi snd typec_ucsi snd_pci_acp3x rng_core fb_sys_fops soundcore libphy rfkill typec battery wmi hp_accel evdev lis3lv02d mac_hid i2c_hid hid acpi_cpufreq hp_wireless ac pinctrl_amd vboxnetflt(OE) vboxnetadp(OE) vboxdrv(OE) drm msr pkcs8_key_parser crypto_user agpgart ip_tables x_tables ext4 crc32c_generic crc16 mbcache jbd2
systemctl status iwd:
● iwd.service - Wireless service
Loaded: loaded (/usr/lib/systemd/system/iwd.service; enabled; vendor preset: disabled)
Active: active (running) since Sun 2020-08-09 20:15:59 CST; 38min ago
Main PID: 377 (iwd)
Tasks: 1 (limit: 18469)
Memory: 3.4M
CGroup: /system.slice/iwd.service
└─377 /usr/lib/iwd/iwd
8月 09 20:16:02 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
8月 09 20:22:54 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
8月 09 20:29:45 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
8月 09 20:29:45 guojize-zhan66-archlinux iwd[377]: Received error during CMD_TRIGGER_SCAN: Network is down (100)
8月 09 20:33:00 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 4, from_ap: false
8月 09 20:36:37 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
8月 09 20:36:37 guojize-zhan66-archlinux iwd[377]: Received error during CMD_TRIGGER_SCAN: Network is down (100)
8月 09 20:43:27 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
8月 09 20:43:27 guojize-zhan66-archlinux iwd[377]: Received error during CMD_TRIGGER_SCAN: Network is down (100)
8月 09 20:50:20 guojize-zhan66-archlinux iwd[377]: Received Deauthentication event, reason: 3, from_ap: false
Offline
https://bbs.archlinux.org/viewtopic.php … 5#p1937315
Still a problem and does the LTS kernel improve the situation for you?
Offline
My WIFI sudently stop working on Lenovo Legion 5 15ARH05 at all, maybe after instaling nvidia drivers. But in yours post I realise that you have iwd and after I installed it and start service it starts to works. Thx
I can for compare show my logs:
Linux Lenovo-ArchLinux 5.9.8-arch1-1 #1 SMP PREEMPT Tue, 10 Nov 2020 22:44:11 +0000 x86_64 GNU/Linux
04:00.0 Network controller: Intel Corporation Wi-Fi 6 AX200 (rev 1a)
Subsystem: Intel Corporation Device 0084
Kernel driver in use: iwlwifi
Kernel modules: iwlwifi
[ 2.452222] iwlwifi 0000:04:00.0: enabling device (0000 -> 0002)
[ 2.461511] iwlwifi 0000:04:00.0: Direct firmware load for iwlwifi-cc-a0-56.ucode failed with error -2
[ 2.463178] iwlwifi 0000:04:00.0: api flags index 2 larger than supported by driver
[ 2.463190] iwlwifi 0000:04:00.0: TLV_FW_FSEQ_VERSION: FSEQ Version: 89.3.35.22
[ 2.463193] iwlwifi 0000:04:00.0: Found debug destination: EXTERNAL_DRAM
[ 2.463194] iwlwifi 0000:04:00.0: Found debug configuration: 0
[ 2.463431] iwlwifi 0000:04:00.0: loaded firmware version 55.d9698065.0 cc-a0-55.ucode op_mode iwlmvm
[ 2.758049] iwlwifi 0000:04:00.0: Detected Intel(R) Wi-Fi 6 AX200 160MHz, REV=0x340
[ 2.979846] iwlwifi 0000:04:00.0: base HW address: 6c:6a:77:5e:4c:dc
[ 3.157276] iwlwifi 0000:04:00.0 wlp4s0: renamed from wlan0
● iwd.service - Wireless service
Loaded: loaded (/usr/lib/systemd/system/iwd.service; enabled; vendor preset: disabled)
Active: active (running) since Fri 2020-11-13 21:33:00 CET; 8min ago
Main PID: 5196 (iwd)
Tasks: 1 (limit: 18380)
Memory: 844.0K
CGroup: /system.slice/iwd.service
└─5196 /usr/lib/iwd/iwd
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Wireless daemon version 1.9
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: station: Network configuration is disabled.
lis 13 21:33:00 Lenovo-ArchLinux systemd[1]: Started Wireless service.
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Wiphy: 0, Name: phy0
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Permanent Address: 6c:6a:77:5e:4c:dc
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Bands: 2.4 GHz 5 GHz
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Ciphers: CCMP TKIP BIP
lis 13 21:33:00 Lenovo-ArchLinux iwd[5196]: Supported iftypes: ad-hoc station ap p2p-client p2p-go p2p-device
lis 13 21:33:15 Lenovo-ArchLinux iwd[5196]: Unexpected connection related event -- is another supplicant running?
lis 13 21:33:15 Lenovo-ArchLinux iwd[5196]: Unexpected connection related event -- is another supplicant running?
I have newer version on Linux. Maybe system update will fix the problem.
Offline
I noticed similar problem but I have AC 9260. It seems like uplink speed is affected in the first place.
linux 5.9.8.arch1-1
linux-lts 5.4.77-1
[ruslan@ruslan-workstation ~]$ lspci | grep Intel
09:00.0 Network controller: Intel Corporation Wireless-AC 9260 (rev 29)
I switched to linux-lts and all went back to normal.
Last edited by mruslan (2020-11-16 14:06:28)
Offline
Similar problems here. I have an ax200 card. With the current kernel download speed is normal, but upload speed is extremely slow. It maxes out at 2MiB/s where I should get around 25MiB/s. With linux-lts speeds are normal
I suspect it has something to do with the firmware that gets loaded.
With current kernel 5.9.9 (*), an error gets reported about loading the file containing the firmware, iwlwifi-cc-a0-56.ucode:
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: enabling device (0000 -> 0002)
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: Direct firmware load for iwlwifi-cc-a0-56.ucode failed with error -2
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: api flags index 2 larger than supported by driver
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: TLV_FW_FSEQ_VERSION: FSEQ Version: 89.3.35.22
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: Found debug destination: EXTERNAL_DRAM
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: Found debug configuration: 0
Nov 20 23:29:19 sovereign kernel: iwlwifi 0000:05:00.0: loaded firmware version 55.d9698065.0 cc-a0-55.ucode op_mode iwlmvm
Nov 20 23:29:20 sovereign kernel: iwlwifi 0000:05:00.0: Detected Intel(R) Wi-Fi 6 AX200 160MHz, REV=0x340
Nov 20 23:29:20 sovereign kernel: iwlwifi 0000:05:00.0: base HW address: 74:d8:xx:xx:xx:xx
Nov 20 23:29:20 sovereign kernel: iwlwifi 0000:05:00.0: iwlmvm doesn't allow to disable BT Coex, check bt_coex_active module parameter
With linux-lts, no such error gets reported, and a different version of the firmware appears to be loaded as well:
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: enabling device (0000 -> 0002)
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: TLV_FW_FSEQ_VERSION: FSEQ Version: 58.3.35.22
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: Found debug destination: EXTERNAL_DRAM
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: Found debug configuration: 0
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: loaded firmware version 50.3e391d3e.0 op_mode iwlmvm
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: Detected Intel(R) Wi-Fi 6 AX200 160MHz, REV=0x340
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: Applying debug destination EXTERNAL_DRAM
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: Allocated 0x00400000 bytes for firmware monitor.
Nov 20 23:32:16 sovereign kernel: iwlwifi 0000:05:00.0: base HW address: 74:d8:xx:xx:xx:xx
Nov 20 23:32:18 sovereign kernel: iwlwifi 0000:05:00.0: Applying debug destination EXTERNAL_DRAM
Nov 20 23:32:18 sovereign kernel: iwlwifi 0000:05:00.0: FW already configured (0) - re-configuring
(*) I'm singling out 5.9.9 here, but I do believe the problem has been going on for a while and affects many recent kernel versions. Grepping through my journald logs, I found that the firmware loading error occurred the first time on August 15th:
Aug 15 15:54:06 sovereign kernel: iwlwifi 0000:05:00.0: Direct firmware load for iwlwifi-cc-a0-56.ucode failed with error -2
On that same date, a few minutes earlier, I find the following entry in my pacman.log:
[2020-08-15T15:50:22+0200] [ALPM] upgraded linux (5.7.12.arch1-1 -> 5.8.1.arch1-1)
Offline
-56 doesn't exist, that's not necessarily a problem. You end up loading -55 BUT on the LTS version you load -50
=> Try to move away /usr/lib/firmware/iwlwifi-cc-a0-55.ucode and reboot the main kernel.
Do NOT delete it! unless you've the LTS kernel or a local copy of linux-firmware or other means to access the internet.
Iff the kernel would only accept -55 you might fail to get the network going without it and have to re-establish its availability.
Offline
-56 doesn't exist, that's not necessarily a problem. You end up loading -55 BUT on the LTS version you load -50
=> Try to move away /usr/lib/firmware/iwlwifi-cc-a0-55.ucode and reboot the main kernel.
Do NOT delete it! unless you've the LTS kernel or a local copy of linux-firmware or other means to access the internet.
Iff the kernel would only accept -55 you might fail to get the network going without it and have to re-establish its availability.
I moved the -55 firmware out of the way, and then it loaded -53. This didn't solve the problem => Still 2MiB/s upload.
I moved the -53 firmware out of the way as well. Then it loaded -50, the same as on the LTS kernel => Still 2MiB/s upload.
I moved the -50 firmware out of the way. Then it loaded -48 => Upload speed now fluctuates between 2MiB/s and 10MiB/s. Better, but still less than half of what I should get, and do get with linux-lts.
Finally I moved the -48 firmware out of the way. Then it loaded -46 => Upload speed still fluctuates between 2MiB/s and 10MiB/s.
So it seems the firmware does have *some* influence, but it's not the real cause.
Offline
Please try downgrading using the ALA linux to https://archive.archlinux.org/packages/ … kg.tar.zst if that is not affected work forwards to the first 5.6 release and so on. To determine which mainline release introduced the issue.
Offline
Please try downgrading using the ALA linux to https://archive.archlinux.org/packages/ … kg.tar.zst if that is not affected work forwards to the first 5.6 release and so on. To determine which mainline release introduced the issue.
So I tested upload speed with various kernels. Here are my results:
5.5.0 -> firmware 50 -> consistently above 20Mib/s
5.5.13 -> firmware 50 -> consistently above 20MiB/s
5.6.0 -> firmware 50 -> no network at all
5.6.2 -> firmware 50 -> fluctuates 10-20MiB/s
5.6.15 -> firmware 50 -> fluctuates 10-20MiB/s
5.7.0 -> firmware 53 -> fluctuates 10-15 MiB/s
5.8.0 -> firmware 55 -> fluctuates 10-15 MiB/s
5.8.15 -> firmware 55 -> fluctuates 8-12 Mib/s
5.9.0 -> firmware 55 -> maxes out at 2 MiB/s
5.9.9 -> firmware 55 -> maxes out at 2 MiB/s
So it seems it was fine up until 5.5.13. Then there was a complete breakage in 5.6.0 and from 5.6.2 on upload speeds got progressively worse and took a big nosedive in 5.9.
Edit to add: 5.6 breakage was apparently caused by this: https://www.phoronix.com/scan.php?page= … el-IWLWIFI
Last edited by BlackMastermind (2020-11-22 14:53:14)
Offline
https://www.phoronix.com/scan.php?page= … el-IWLWIFI
https://www.phoronix.com/scan.php?page= … 2-Released
kr00k mitigation efforts?
Edit: can you try an unprotected connection?
Edit #2: I need to F5 more often…
Last edited by seth (2020-11-22 15:15:01)
Offline
So I did some more googling, and found this:
https://bugzilla.kernel.org/show_bug.cgi?id=209913#c4
I tested the solution mentioned in that thread, and it seems to work. The upload speed isn't as stable as in 5.5 and earlier, sometimes it drops well below 20MiB/s for a few seconds, and then picks up again, but it's certainly much more usable than before.
tl;dr: turn TSO off to restore upload speed with this command:
sudo ethtool -K wlp3s0 tso off
Last edited by BlackMastermind (2020-11-22 17:02:48)
Offline
Can confirm that turning off TCP segmentation offload fixes the problem for me as well on 5.9.10. LTS kernel (5.4.79) works fine without the fix.
I have added the following file to /etc/systemd/network to make it persistent:
# /etc/systemd/network/10-intel-wifi.link
[Match]
Driver=iwlwifi
[Link]
TCPSegmentationOffload=false
NamePolicy=keep kernel database onboard slot path
The same can be achieved using a udev rule if not using systemd.
Offline
I'm getting painfully slow download speeds with an AX201 card on a Lenovo X1 Carbon. Sometimes after a reboot performance returns
to normal, but after a short time I'm getting 30-50kbps down.
Pings are around 5s to 8.8.8.8 with 63% packet loss.
Disabling TSO makes zero difference to the speed. I'm very patiently waiting for the lts kernel to download so I can test it.
Edit: nope, no difference. I'll open a new thread.
Last edited by bobthemighty (2020-12-05 13:35:41)
Offline
For anybody interested in udev solution:
cat > /etc/udev/rules.d/95-iwlwifi-ax200-no-tso.rules << EOF
# name of this rule file should sort quite late, specifically after persistent naming rules
# because of how substitutions and device renames interact
# some change in 5.9 breaks TCP TSO on iwlwifi
# see https://bbs.archlinux.org/viewtopic.php?id=258042
# upsteam bug https://bugzilla.kernel.org/show_bug.cgi?id=209913
ACTION=="add", SUBSYSTEM=="net", DRIVERS=="iwlwifi", ATTRS{vendor}=="0x8086", ATTRS{device}=="0x2723", RUN+="/usr/sbin/ethtool -K \$name tso off"
EOF
This is specifically for AX200 on Fedora (ethtool path is different).
Last edited by jmennius (2020-12-14 08:40:22)
Offline
I'm having a similar issue with TSO on my Intel AC 9560 on kernel 5.9.12. Only TX (upload) is affected, RX is fine. The speed has dropped significantly, but I've also noticed a problem with SMB shares, where every upload results in this:
$ smbclient --no-pass --command='put random.bin' //samba/share
cli_push returned NT_STATUS_IO_TIMEOUT
NT_STATUS_CONNECTION_DISCONNECTED closing remote file \random.bin
Network traffic captured on the Samba server shows some strange/malformed TCP headers coming from the client. Turning the TCP segmentation offload off fixes both issues.
Offline
I have the same issue with the RTL8812AE 802.11ac PCIe Wireless Network Adapter. On 5.10.3 the WiFi upload speed is very slow but if I switch back to 5.4 LTS the WiFi speed is back to normal. Running:
sudo ethtool -K wlp2s0 tso off
didn't change anything. Anyone else found a solution? Does upstream know about this issue and that it isn't only affecting Intel cards?
Offline
I still have this problem with AX201. Switching off tso seems to improve the speed but the wifi keeps disconnects/reconnects by NetworkManager..
Offline
Frequent reconnects can also be due to concurrent control services.
find /etc/systemd -type l -exec test -f {} \; -print | awk -F'/' '{ printf ("%-40s | %s\n", $(NF-0), $(NF-1)) }' | sort -f
Otherwise NM causes this for https://wiki.archlinux.org/index.php/Ne … domization - so disable that.
If none of this helps, post a complete system jorunal covering multiple reconnects.
Edit: lingo
Last edited by seth (2021-01-11 08:42:30)
Offline
Thanks for this! Looks like to disable mac randomization works. It seems each reconnect ion needs to disable TSO another time to bring back the speed.. so these two things convoluted together..
Offline
In case others might find it useful, using iw backend of NetworkManager instead of wpa_supplicant solves all the problems. Need to mark the wpa_supplicant after changing to iw backend in case NM starts it automatically `systemctl mask wpa_supplicant.service`
Offline
Pages: 1