lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
Open Source and information security mailing list archives
| ||
|
Message-ID: <c21206d6-eb4b-2e7d-eb2e-cab9f3b06c62@lwfinger.net> Date: Wed, 10 May 2023 21:45:37 -0500 From: Larry Finger <Larry.Finger@...inger.net> To: Yun Lu <luyun_611@....com>, Bitterblue Smith <rtl8821cerfe2@...il.com> Cc: Jes.Sorensen@...il.com, kvalo@...nel.org, davem@...emloft.net, edumazet@...gle.com, kuba@...nel.org, pabeni@...hat.com, linux-wireless@...r.kernel.org, netdev@...r.kernel.org Subject: Re: [PATCH] wifi: rtl8xxxu: fix authentication timeout due to incorrect RCR value On 5/10/23 21:29, Yun Lu wrote: > Larry and Bitterblue: > > Thank you for your reply, are there any further questions or suggestions on this issue? > Could this patch be merged? There seems to be no other side effects. > > > > 在 2023-05-04 15:39:57,"Yun Lu" <luyun_611@....com> 写道: >> >> 在 2023-04-30 18:36:50,"Bitterblue Smith" <rtl8821cerfe2@...il.com> 写道: >>> On 29/04/2023 06:35, Yun Lu wrote: >>>> At 2023-04-29 01:06:03, "Larry Finger" <Larry.Finger@...inger.net> wrote: >>>>> On 4/27/23 23:11, wo wrote: >>>>>> [ 149.595642] [pid:7,cpu6,kworker/u16:0,0]BEFORE: REG_RCR differs from regrcr: >>>>>> 0x1830613 insted of 0x7000604e >>>>>> [ 160.676422] [pid:237,cpu6,kworker/u16:5,3]BEFORE: REG_RCR differs from >>>>>> regrcr: 0x70006009 insted of 0x700060ce >>>>>> [ 327.234588] [pid:7,cpu7,kworker/u16:0,5]BEFORE: REG_RCR differs from >>>>> regrcr: 0x1830d33 insted of 0x7000604e >>>>> >>>>> >>>>> My patch was messed up, but it got the information that I wanted, which is shown >>>>> in the quoted lines above. One of these differs only in the low-order byte, >>>>> while the other 2 are completely different. Strange! >>>>> >>>>> It is possible that there is a firmware error. My system, which does not show >>>>> the problem, reports the following: >>>>> >>>>> [54130.741148] usb 3-6: RTL8192CU rev A (TSMC) romver 0, 2T2R, TX queues 2, >>>>> WiFi=1, BT=0, GPS=0, HI PA=0 >>>>> [54130.741153] usb 3-6: RTL8192CU MAC: xx:xx:xx:xx:xx:xx >>>>> [54130.741155] usb 3-6: rtl8xxxu: Loading firmware rtlwifi/rtl8192cufw_TMSC.bin >>>>> [54130.742301] usb 3-6: Firmware revision 88.2 (signature 0x88c1) >>>>> >>>>> Which firmware does your unit use? >>>> >>>> The firmware verion we used is 80.0 (signature 0x88c1) >>>> [ 903.873107] [pid:14,cpu0,kworker/0:1,2]usb 1-1.2: RTL8192CU rev A (TSMC) 2T2R, TX queues 2, WiFi=1, BT=0, GPS=0, HI PA=0 >>>> [ 903.873138] [pid:14,cpu0,kworker/0:1,3]usb 1-1.2: RTL8192CU MAC: 08:be:xx:xx:xx:xx >>>> [ 903.873138] [pid:14,cpu0,kworker/0:1,4]usb 1-1.2: rtl8xxxu: Loading firmware rtlwifi/rtl8192cufw_TMSC.bin >>>> [ 903.873474] [pid:14,cpu0,kworker/0:1,5]usb 1-1.2: Firmware revision 80.0 (signature 0x88c1) >>>> >>>>> >>>>> Attached is a new test patch. When it logs a CORRUPTED value, I would like to >>>>> know what task is attached to the pid listed in the message. Note that the two >>>>> instances where the entire word was wrong came from pid:7. >>>>> >>>>> Could improper locking could produce these results? >>>>> >>>>> Larry >>>> >>>> Apply your new patch, then turn on/off the wireless network switch on the network control panel serverl loops. >>>> The log shows: >>>> [ 85.384429] [pid:221,cpu6,kworker/u16:6,5]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x70006009 insted of 0x700060ce >>>> [ 121.681976] [pid:216,cpu6,kworker/u16:3,0]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x70006009 insted of 0x700060ce >>>> [ 144.416992] [pid:217,cpu6,kworker/u16:4,1]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x70006009 insted of 0x700060ce >>>> >>>> And if we up/down the interface serverl loops as follows: >>>> ifconfig wlx08bexxxxxx down >>>> sleep 1 >>>> ifconfig wlx08bexxxxxx up >>>> sleep 10 >>>> The log shows: >>>> [ 282.112335] [2023:04:29 10:30:34][pid:95,cpu6,kworker/u16:1,3]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x1832e13 insted of 0x7000604e >>>> [ 293.311462] [2023:04:29 10:30:45][pid:217,cpu7,kworker/u16:4,9]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x1830e72 insted of 0x7000604e >>>> [ 304.435089] [2023:04:29 10:30:56][pid:217,cpu6,kworker/u16:4,9]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x1830ed3 insted of 0x7000604e >>>> [ 315.532257] [2023:04:29 10:31:07][pid:95,cpu7,kworker/u16:1,8]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x7000604e insted of 0x7000604e >>>> [ 324.114379] [2023:04:29 10:31:16][pid:221,cpu6,kworker/u16:6,7]REG_RCR corrupted in rtl8xxxu_configure_filter: 0x1832e14 insted of 0x7000604e >>>> >>>> We also update the firmware verion to 88.2, and the test results are the same as above. >>>> >>>> Thank you for helping debug this issue, which seems to be related to specific devices. >>>> >>>> Yun Lu >>>> >>>> >>>> >>>> >>> There was this bug report about phantom MAC addresses with >>> the RTL8188CUS: >>> https://lore.kernel.org/linux-wireless/a31d9500-73a3-f890-bebd-d0a4014f87da@reto-schneider.ch/ >>> >>> See the pcap file. I wonder if it's related? >> >> The bug in the link is a high retransmission rate during message transmission, but the problem we encountered is that >> the nic cannot receive authentication frames, resulting in authentication timeout and inability to connect to WiFi. It seems >> that these two issues are not related. >> >> We also enabled monitor mode and found that the AP has replied to the authentication message, but the nic cannot receive >> this reply message due to the incorrect RCR register value. Once the RCR register is modified to the correct value, >> the authentication message can be received normally and the connection to WIFI can be normal. Yun Lu, I have no objection to adding this patch. Although it looked a little ad-hoc at first, it seems to fix a hardware or firmware error for your device. It certainly does no harm other than taking up a bit of memory in the loaded driver. Resubmit the patch with a new version number, and I will Ack it. Larry Larry
Powered by blists - more mailing lists