Closed csirikak closed 7 years ago
From: pmcduff [mailto:notifications@github.com] Sent: Sunday, September 24, 2017 2:02 PM To: kaloz/mwlwifi Cc: Subscribed Subject: [kaloz/mwlwifi] Wrt3200acm freezes up. (#217)
I reverted back to dd-wrt as the stock firmware was too unstable. I got these crash logs after about 4 mins of > uptime. It causes all clients to lose internet connectivity, and the router becomes unpingable. [ 237.961237] INFO: rcu_sched self-detected stall on CPU
Again
[ 237.966411] 1-...: (1 GPs behind) idle=173/140000000000001/0 softirq=6360/6361 fqs=3000
[ 237.971238] INFO: rcu_sched detected stalls on CPUs/tasks:
[ 237.971242] 1-...: (1 GPs behind) idle=173/140000000000001/0 softirq=6360/6361 fqs=3000
[ 237.971245] (detected by 0, t=6002 jiffies, g=2501, c=2500, q=891)
[ 237.971246] Task dump for CPU 1:
[ 237.971250] kworker/u4:1 R running task 0 17 2 0x00000002
[ 237.971263] Workqueue: phy1 _113 [mwlwifi]
[ 237.971272] [
[ 238.034703] [
RCU stall is fixed by 0810.
This is the driver that has problem with 'Stall On CPU' how is 0810 fixed????
driver name: mwlwifi chip type: 88W8964 hw version: 7 driver version: 10.3.4.0-20170810 firmware version: 0x09030007 power table loaded from dts: no firmware region code: 0x10 mac address: 60:38:e0:13:64:f9 2g: enable 5g: disable antenna: 4 4 irq number: 47 ap macid support: 0000ffff sta macid support: 00010000 macid used: 00000001 radio: enable iobase0: e1000000 iobase1: e1280000 tx limit: 1024 rx limit: 16384
0810 is not fixed for CPU STALL!
You got this problem?
0913 LEDE main trunk still uses 0713.
Why are we biased against dd-wrt, OpenWRT/LEDE is not a perfect project nor is it easily configurable.
Is LEDE sponsoring this driver or is Belkin pushing for support mainly for LEDE?
On Sep 24, 2017 7:19 PM, "yuhhaurlin" notifications@github.com wrote:
0913 LEDE main trunk still uses 0713.
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/kaloz/mwlwifi/issues/217#issuecomment-331750132, or mute the thread https://github.com/notifications/unsubscribe-auth/ARIphH9abuNhSKwI-wrO1TpMXiLSL5mqks5slvGpgaJpZM4PiAnp .
047] ieee80211 phy0: staid 1 deleted
[ 8535.148588] ieee80211 phy0: staid 1 deleted
[ 8567.554564] ieee80211 phy1: staid 2 deleted
[ 8618.614571] ieee80211 phy1: staid 3 deleted
[ 8629.416566] ieee80211 phy1: staid 3 deleted
[ 8638.378584] ieee80211 phy1: staid 3 deleted
[ 8700.748423] ieee80211 phy0: cmd 0x801d=MEMAddrAccess timed out
[ 8700.754294] ieee80211 phy0: return code: 0x051d
[ 8700.758844] ieee80211 phy0: timeout: 0x001d
[ 8704.762414] ieee80211 phy0: cmd 0x801d=MEMAddrAccess timed out
[ 8704.768275] ieee80211 phy0: return code: 0x001d
[ 8704.772833] ieee80211 phy0: timeout: 0x001d
[ 8708.776421] ieee80211 phy0: cmd 0x801d=MEMAddrAccess timed out
[ 8708.782289] ieee80211 phy0: return code: 0x001d
[ 8708.786838] ieee80211 phy0: timeout: 0x001d
[ 8839.210401] ieee80211 phy0: cmd 0x9143=GetSeqno timed out
[ 8839.215836] ieee80211 phy0: return code: 0x1143
[ 8839.220409] ieee80211 phy0: timeout: 0x1143
[ 9138.180444] ieee80211 phy0: can't send packet during ADDBA
[ 9142.190238] Unable to handle kernel NULL pointer dereference at virtual address 0000000c
[ 9142.198380] pgd = d4cc0000
[ 9142.201105] [0000000c] pgd=12f1b831, pte=00000000, *ppte=00000000
[ 9142.207431] Internal error: Oops: 17 [#1] SMP ARM
[ 9142.212153] Modules linked in: nf_nat_pptp nf_conntrack_pptp nf_nat_proto_gre nf_conntrack_proto_gre mip6 ip6_tunnel xfrm6_mode_beet xfrm6_mode_transport xfrm6_mode_tunnel xfrm6_tunnel ipv6 tunnel6 fuse ledtrig_usbport pwrseq_simple pwrseq_emmc ahci_mvebu ahci_platform ahci libahci_platform libahci ehci_orion sata_mv usb_storage sd_mod xhci_plat_hcd xhci_pci xhci_hcd ehci_pci ehci_platform ehci_hcd usbcore usb_common btmrvl_sdio btmrvl bluetooth mwifiex_sdio mvsdio sdhci_pxav3 sdhci_pltfm sdhci mmc_block mmc_core mwifiex mwlwifi mac80211 cfg80211 compat mii tmp421 pwm_fan leds_pca963x leds_tlc591xx orion_wdt [last unloaded: nf_conntrack_proto_gre]
[ 9142.270300] CPU: 0 PID: 2429 Comm: hostapd Not tainted 4.9.50 #51
[ 9142.276418] Hardware name: Marvell Armada 380/385 (Device Tree)
[ 9142.282362] task: df4e0540 task.stack: d59b6000
[ 9142.286918] PC is at _145+0x98/0xf0 [mwlwifi]
[ 9142.291295] LR is at _145+0x34/0xf0 [mwlwifi]
[ 9142.295669] pc : [
@rs-se For LEDE with updated mwlwifi driver, please check https://davidc502sis.dynamic-dns.net/,
@aaron1959 This is still happened on dd-WRT, right? Please reproduce it on LEDE via https://davidc502sis.dynamic-dns.net/. Let me know the LEDE image you used and the way to reproduce the problem. I will hook ICE to check this problem.
@rs-se I don't know who takes care of updating mwlwifi with LEDE. But link here https://davidc502sis.dynamic-dns.net/ will keep updated mwlwifi driver with LEDE image for WRT series devices.
No need for LEDE the needed info is already posted above. No need to hook ICE to check.
@aaron1959 This is host command time out. Please follow the way I told you to reproduce the problem. So I can check the problem. Otherwise, I can't help you.
@yuhhaurlin i already use this newer firmware in dd-wrt and posted it in our forum for testing. i extracted it from the linksys firmware. so far no new issues. some people say its somewhat faster. but the reboot / crash problems remain with certain clients
@BrainSlayer Can you try to load the image I posted (9.3.0.8) to make sure it is all right? If yes, I will post the firmware to this GitHub. Thanks.
@BrainSlayer Another strange thing is that why they can't reproduce this problem on LEDE image?
@yuhhaurlin the problem is see is that right now all people here avoid to use lede for some reason. all people who are posting about this issue are using mainly dd-wrt. i just can guarantee that i use the plain unmodified mwlwifi driver from github here using a 4.9 kernel. the kernel patches required for the wrt3200acm is based on lede sources. but however i may use different hostapd configs. so i may use hostapd vht flags which arent enabled in lede
@BrainSlayer Capability will be added to hostapd configuration file via script file. I wonder what kind of VHT capability won't be added to configuration file of hostapd?
these are my capabilities with vht160
ht_capab=[HT40-][LDPC][SHORT-GI-20][SHORT-GI-40][DSSS_CCK-40][MAX-AMSDU-7935] vht_capab=[RXLDPC][SHORT-GI-80][SHORT-GI-160][RX-STBC-1][RX-ANTENNA-PATTERN][TX-ANTENNA-PATTERN][MAX-MPDU-11454][VHT160][MAX-A-MPDU-LEN-EXP7]
i read out the supported caps from the mwlwifi driver and generated it in that way.
I think LEDE do the same thing, right?
Script files will do this kind of thing. In order to support some features, I had patched mac80211.sh and hostpad.sh. And I think these script files will add capability of driver to hostapd configuration file.
if i remember correct lede does not support all of this flags. can you check your own flags and compare it with mine? or simply post it here and i compare it
so the script in lede supports only features known by the script itself.
I don't have 88W8964 module now. Can anyone help to post hostapd configuration file of WRT3200ACM here (/var/run/hostapd-phy0.conf and /var/run/hostapd-phy1.conf, one is 2g and another one is 5g).
LEDE is fine - I weekly build my own latest version - no crashes - no issues with WRT3200ACM
DD-WRT is basically behind support - hostpad, linux kernel etc
this is a red herring basically
@rmandrad Can you help to dump hostapd configuration file? Thanks.
let me know if this is what you want ....
hostapd-phy0.conf
driver=nl80211 logger_syslog=127 logger_syslog_level=2 logger_stdout=127 logger_stdout_level=2 country_code=FR ieee80211d=1 ieee80211h=1 hw_mode=a beacon_int=100 channel=124
ieee80211n=1 ht_coex=0 ht_capab=[HT40+][LDPC][SHORT-GI-20][SHORT-GI-40][MAX-AMSDU-7935][DSSS_CCK-40] vht_oper_chwidth=2 vht_oper_centr_freq_seg0_idx=114 ieee80211ac=1 vht_capab=[RXLDPC][SHORT-GI-80][SHORT-GI-160][SU-BEAMFORMER][SU-BEAMFORMEE][RX-ANTENNA-PATTERN][TX-ANTENNA-PATTERN][RX-STBC-1][VHT160][MAX-MPDU-11454][MAX-A-MPDU-LEN-EXP7]
interface=wlan0 ctrl_interface=/var/run/hostapd ap_isolate=1 disassoc_low_ack=1 preamble=1 wmm_enabled=1 ignore_broadcast_ssid=0 uapsd_advertisement_enabled=1 wpa_passphrase=xxx wpa=2 wpa_pairwise=CCMP ssid=xx bridge=br-lan wpa_key_mgmt=WPA-PSK okc=0 disable_pmksa_caching=1 bssid=xxxx
note that I removed phy2 ... the phy1 runs on 2g
@rmandrad Yes. Thanks.
@BrainSlayer I think script files of LEDE completely add capability of mwlwifi driver to hostapd configuration file. And this is what I think.
its identical. except that in dd-wrt Single user beamforming is disabled by default and just enabled if user enables it in gui
@rmandrad : dd-wrt uses kernel 4.9. so what you say is not correct. its always latest. for hostapd i use a little bit older versions since there was no issue right now which forces my to use a newer version
but phy2 is enabled by default in dd-wrt. but can be disabled
which version of hostapad does DD-WRT run ? I remember that LEDE had an issue with hostapad on versions earlier than 2017... suggest you upgrade to latest hostapad
i dont have issues. its october 2016 i use. it was originally takes from openwrt. but i avoid updates without a reason since hostapd growes by codesize alot each time which causes troubles for smaller routers
but hostapd is unlikelly to cause crashes in the kernel. if so, its still a driver bug
well ... I don't know DD-WRT ...... compiling LEDE I found some issue with GCC vs MUSL... so I use MUSL
you mean uclibc or glibc. dd-wrt uses musl too
Yes musl works fine not gcc .. don't know uclibc
On 25 Sep 2017 12:47, "Sebastian Gottschall" notifications@github.com wrote:
you mean uclibc or glibc. dd-wrt uses musl too
— You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub https://github.com/kaloz/mwlwifi/issues/217#issuecomment-331856744, or mute the thread https://github.com/notifications/unsubscribe-auth/AFHz7NyI5S_SZkCEMkxgavKHvT4FABtrks5sl5LOgaJpZM4PiAnp .
gcc is no library. gcc is a compiler which uses uclibc, glibc or musl as backend c library
sorry .. what I am saying it may be worth compiling dd-wrt with musl ...
@BrainSlayer, Here's the hostapd conf from dd-wrt: ath0:
ieee80211n=1
dynamic_ht40=0
ht_capab=[HT40+][LDPC][SHORT-GI-20][SHORT-GI-40][DSSS_CCK-40][MAX-AMSDU-7935]
vht_capab=[RXLDPC][SHORT-GI-80][RX-STBC-1][SU-BEAMFORMER][RX-ANTENNA-PATTERN][TX-ANTENNA-PATTERN][MAX-MPDU-11454][MAX-A-MPDU-LEN-EXP7][BF-ANTENNA-2][SOUNDING-DIMENSION-2][BF-ANTENNA-3][SOUNDING-DIMENSION-3]
ieee80211ac=1
vht_oper_chwidth=1
vht_oper_centr_freq_seg0_idx=42
hw_mode=a
channel=36
frequency=5180
beacon_int=100
ath1:
ieee80211n=1
dynamic_ht40=0
ht_capab=[HT20][LDPC][SHORT-GI-20][SHORT-GI-40][DSSS_CCK-40][MAX-AMSDU-7935]
vht_capab=[RXLDPC][RX-STBC-1][SU-BEAMFORMER][RX-ANTENNA-PATTERN][TX-ANTENNA-PATTERN][MAX-MPDU-11454][MAX-A-MPDU-LEN-EXP7][BF-ANTENNA-2][SOUNDING-DIMENSION-2][BF-ANTENNA-3][SOUNDING-DIMENSION-3]
vendor_vht=1
hw_mode=g
channel=9
frequency=2452
beacon_int=100
Note the addition of [SOUNDING-DIMENSION-2] and [BF-ANTENNA-3]
@rmandrad and all what i say is that dd-wrt is using musl since more than 2 years if not longer
@pmcduff yes thats the configs i mean which do not exist in openwrt. you found them. thanks. @yuhhaurlin any oppinions on these options which arent supported by lede?
@yuhhaurlin the only main difference is that dd-wrt makes use of single user beamforming if selected. it can be disabled in wireless settings. but if this is causing the big fail i may exlude this feature from mwlwifi until its fixes. @pmcduff you can disable single user beamforming at wireless basic advanced settings. try if its more stable then
@BrainSlayer DD-WRT single user beam forming is disabled by default.
What really helped was to change txpower from 30dBm to 27, it made a huge difference. I'm getting about an hour of stable operation as opposed to a few minutes even with single-user beamforming enabled. I will try disabling it to see if that helps even more, but it is a nice feature that I would not like to disable. DD-WRT v3.0-r33006 std(08/03/17)
Tx power is not supported.
It crashed after 20 minutes with single-user beamforming disabled.
I reverted back to dd-wrt as the stock firmware was too unstable. I got these crash logs after about 4 mins of uptime. It causes all clients to lose internet connectivity, and the router becomes unpingable.