The router is beside me, but weak signal, how to debug?
I know nothing of the SDIO card or driver. If you can get access to the card, does it have one or two antennas connected? If only one, try moving the antenna to the other connector. Does that help?
(In reply to Larry Finger from comment #1) > I know nothing of the SDIO card or driver. If you can get access to the > card, does it have one or two antennas connected? If only one, try moving > the antenna to the other connector. Does that help? from drivers/staging/rtl8723bs source code, i need add which one to cc list help me?
The router is beside me,on 4.14.78 signal is well,connect wifi signal well. # iw dev wlan0 scan | egrep "SSID|signal" signal: -45.00 dBm SSID: YOULING on 4.19/4.18/4.17 kernel,connect wifi,the router is beside me,signal final 90dbm. # iw dev wlan0 scan | egrep "SSID|signal" signal: -78.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -83.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -83.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -85.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -85.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -87.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -88.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -90.00 dBm SSID: YOULING # iw dev wlan0 scan | egrep "SSID|signal" signal: -90.00 dBm SSID: YOULING
I have no idea why the signal varies so much from scan to scan unless you are in a place with multiple APs all broadcasting the same SSID. To test that, change the command to sudo iw dev wlp0s20u6 scan | egrep "SSID|signal|\(on" Post the results.
(In reply to Larry Finger from comment #4) > I have no idea why the signal varies so much from scan to scan unless you > are in a place with multiple APs all broadcasting the same SSID. To test > that, change the command to > > sudo iw dev wlp0s20u6 scan | egrep "SSID|signal|\(on" > > Post the results. signal Weaker and weaker,can you understand? only one ssid, router is beside me only 50cm.
(In reply to Larry Finger from comment #4) > 我不知道为什么信号在扫描到扫描之间变化很大，除非你在一个有多个AP的地方都广播相同的SSID。要测试它，请将命令更改为 > > sudo iw dev wlp0s20u6 scan | egrep“SSID | signal | \（on” > > 发布结果。 connect ssid 10 second,signal Weaker and weaker,final 95 -dbm,router is beside me only 50cm.
Then there may be nothing I can do. The strength of the AP is set by the firmware in the AP, not by the driver in the station. As you have seen -45 dBm, that means that the driver is working OK, at least once. One possibility is that the station is too strong. Try moving the computer to a distance of about 2m and try again. Is there a possibility of trying a different AP?
I too have the same issue, except for the disconnect part. The wifi signal on rtl8723bs on 4.19rc7 show good strength before connection, but as soon as the connection is established the strength reduces to less than 30% or so, however it doesn't affect the speed or performance. The same router and setup work fine (signal strength > 90%) when I reboot and select my 4.14+ kernel. Config file for 4.19rc7: https://gist.github.com/srikantpatnaik/0d78b23345e9a36d022a1139fac0f70c Dmesg output for 4.19rc7: https://gist.github.com/srikantpatnaik/deecd8adf561e2b5704346aa82ecfe6c Is it possible to get the correct signal strength on this kernel?
Yes, but someone with the problem will need to bisect between 4.14 and 4.19 to see what commit causes the problem. There are a number of non-trivial changes in kernel 4.17, but I would not know where to look.
checkout df34df483a97b1591a3e90a6941f99fe9f863508, test single low, checkout 9abf8acea297b4c65f5fa3206e2b8e468e730e84, test single normal, 582 patches，37 patches for rtl8723bs, git reset --hard 9abf8acea297b4c65f5fa3206e2b8e468e730e84, git am these 37 patches, test single low. now, find problem on my branch, $ git format-patch -s 6e5750427e71b729ed46d777e28915adc28d5b35 0001-staging-rtl8723bs-make-myid-function-to-follow-kerne.patch 0002-staging-rtl8723bs-clean-up-conditionals.patch 0003-staging-rtl8723bs-Remove-unneeded-cast.patch 0004-staging-rtl8723bs-use-kmemdup-for-allocation-and-cop.patch 0005-staging-rtl8723bs-fix-u8-less-than-zero-check.patch 0006-staging-rtl8723bs-Remove-unnecessary-semicolon.patch 0007-staging-rtl8723bs-Replace-memset-with-eth_zero_addr.patch 0008-staging-rtl8723bs-core-rtw_cmd-remove-unnecessary-in.patch 0009-Staging-rtl8723bs-static-variables-are-always-0.patch 0010-Staging-rtl8723bs-Remove-unnecessary-braces.patch 0011-Staging-rtl8723bs-clean-up-spaces.patch 0012-Staging-rtl8723bs-Add-blank-line-after-declaration.patch 0013-staging-rtl8723bs-Remove-duplicate-defines.patch 0014-staging-rtl8723bs-Remove-defines-shadowing-enums-in-.patch 0015-staging-rtl8723bs-Replace-RTW_IEEE80211_FCTL_-with-I.patch 0016-staging-rtl8723bs-Replace-RTW_IEEE80211_FTYPE_-with-.patch 0017-staging-rtl8723bs-Replace-RTW_IEEE80211_STYPE_-with-.patch 0018-staging-rtl8723bs-Fix-newlines-in-rtw_wx_set_auth.patch 0019-staging-rtl8723bs-Remove-unecessary-braces-from-swit.patch 0020-staging-rtl8723bs-Remove-braces-from-single-statemen.patch 0021-staging-rtl8723bs-Fix-alignment-in-rtw_wx_set_auth.patch 0022-staging-rtl8723bs-Fix-IEEE80211-authentication-algor.patch 0023-staging-rtl8723bs-Remove-unnecessary-length-define-s.patch 0024-staging-rtl8723bs-Fix-lines-with-trailing-open-paren.patch 0025-staging-rtl8723bs-Add-spaces-around-ternary-operator.patch 0026-staging-rtl8723bs-Add-missing-braces-in-else-stateme.patch 0027-staging-rtl8723bs-Change-camel-case-to-snake-case-in.patch 0028-staging-rtl8723bs-Remove-unnecessary-blank-lines-in-.patch 0029-staging-rtl8723bs-Fix-lines-too-long-in-update_recvf.patch 0030-staging-rtl8723bs-Fix-function-signature-that-goes-o.patch 0031-staging-rtl8723bs-Factor-out-rtl8723bs_recv_tasklet-.patch 0032-staging-rtl8723bs-Replace-NULL-pointer-comparison-wi.patch 0033-staging-rtl8723bs-Rework-struct-_ODM_Per_Pkt_Info_-c.patch 0034-staging-rtl8723bs-Rework-struct-_ODM_Phy_Status_Info.patch 0035-staging-rtl8723bs-Remove-unecessary-newlines-from-od.patch 0036-staging-rtl8723bs-Replace-yield-call-with-cond_resch.patch 0037-staging-rtl8723bs-Remove-yield-call-replace-with-con.patch $ git reset --hard 0a84d43431a27898b9383cec84d34bab10b437ea HEAD is now at 0a84d43431a2 staging: rtl8723bs: Remove #defines shadowing enums in 'linux/ieee80211.h' (single normal) $ git reset --hard 73550e32363332f3a4a0223bf703c8f19b23f08b HEAD is now at 73550e323633 staging: rtl8723bs: Replace RTW_IEEE80211_STYPE_* with IEEE80211_STYPE_*. (single normal) $ git reset --hard cb45367a3c3ad84ee3e3b8f1b515750126d814d1 HEAD is now at cb45367a3c3a staging: rtl8723bs: Add spaces around ternary operators. (single normal) $ git reset --hard 40e24fe89b8102374d05a50b29531232f4377604 HEAD is now at 40e24fe89b81 staging: rtl8723bs: Fix function signature that goes over 80 characters. (single low) $ git reset --hard 0203a50434e82fa5d85cff8a4d781f4dcdf56680 HEAD is now at 0203a50434e8 staging: rtl8723bs: Change camel case to snake case in 'rtl8723bs_recv.c'. (single normal) $ git reset --hard 2c3c92cb59fa9131c6f46db3afc0bf8d70406530 HEAD is now at 2c3c92cb59fa staging: rtl8723bs: Fix lines too long in update_recvframe_attrib(). (single low) $
Created attachment 279331 [details] revert patch Revert "staging: rtl8723bs: Factor out rtl8723bs_recv_tasklet() sections." Revert "staging: rtl8723bs: Fix function signature that goes over 80 characters." Revert "staging: rtl8723bs: Fix lines too long in update_recvframe_attrib()." signal normal on 4.20 kernel.
Thanks for the testing. Looking at your attachment and evaluating each of the patches: 1. 0001-Revert-staging-rtl8723bs-Replace-NULL-pointer-compar.patch cannot be causing a problem unless you have a gcc problem. 2. 0002-Revert-staging-rtl8723bs-Factor-out-rtl8723bs_recv_t.patch - I have not analyzed this patch. 3. 0003-Revert-staging-rtl8723bs-Fix-function-signature-that.patch looks OK. It just splits a long line. 4. 0004-Revert-staging-rtl8723bs-Fix-lines-too-long-in-updat.patch - This change has a problem: The original code - pkt_info.bssid_match = ((!IsFrameTypeCtrl(wlanhdr)) && - !pattrib->icv_err && !pattrib->crc_err && - !memcmp(get_hdr_bssid(wlanhdr), get_bssid(&padapter->mlmepriv), ETH_ALEN)); was replaced by + pkt_info.bssid_match = ((!IsFrameTypeCtrl(wlanhdr)) && + !pattrib->icv_err && !pattrib->crc_err && + !ether_addr_equal(rx_bssid, my_bssid)); The problem is that memcmp() returns 0 (or false) when the two items are equal, whereas ether_addr_equal() returns true when the items are equal. The following patch will fix that problem: diff --git a/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c b/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c index 85077947b9b8..d2b9d99ac480 100644 --- a/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c +++ b/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c @@ -109,7 +109,7 @@ static void update_recvframe_phyinfo(union recv_frame *precvframe, rx_bssid = get_hdr_bssid(wlanhdr); pkt_info.bssid_match = ((!IsFrameTypeCtrl(wlanhdr)) && !pattrib->icv_err && !pattrib->crc_err && - !ether_addr_equal(rx_bssid, my_bssid)); + ether_addr_equal(rx_bssid, my_bssid)); rx_ra = get_ra(wlanhdr); my_hwaddr = myid(&padapter->eeprompriv); The white space will be corrupted in the above listing, thus you will need to do the edit, but removing one character should not be a problem. :) After the above one character change, does the driver function correctly, or do I need to look harder at #2?
(In reply to Larry Finger from comment #12) > The following patch will fix that problem: > > diff --git a/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c > b/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c > index 85077947b9b8..d2b9d99ac480 100644 > --- a/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c > +++ b/drivers/staging/rtl8723bs/hal/rtl8723bs_recv.c > @@ -109,7 +109,7 @@ static void update_recvframe_phyinfo(union recv_frame > *precvframe, > rx_bssid = get_hdr_bssid(wlanhdr); > pkt_info.bssid_match = ((!IsFrameTypeCtrl(wlanhdr)) && > !pattrib->icv_err && !pattrib->crc_err && > - !ether_addr_equal(rx_bssid, my_bssid)); > + ether_addr_equal(rx_bssid, my_bssid)); > > rx_ra = get_ra(wlanhdr); > my_hwaddr = myid(&padapter->eeprompriv); > > The white space will be corrupted in the above listing, thus you will need > to do the edit, but removing one character should not be a problem. :) > > After the above one character change, does the driver function correctly, or > do I need to look harder at #2? Yes,only the above one character change, signal normal.
Thank you for testing. Do you want to prepare the patch for the kernel, or should I? If the latter, is it OK to list your name and E-mail in a "reported-and-tested-by" line?
(In reply to Larry Finger from comment #14) > Thank you for testing. Do you want to prepare the patch for the kernel, or > should I? If the latter, is it OK to list your name and E-mail in a > "reported-and-tested-by" line? Reported-and-tested-by: youling257 <youling257@gmail.com>
I need a first and last name, not a "screen" name. That name can be an anglicized form if that is what you want, or a formal name.
Reported-and-tested-by, needn't real name. https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c0d730885e6e5180cbb593f25babffa878e18fef https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=44960f2a7b63e224b1091b3e1d6f60e0cdf4be0c https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=1b3a62643660020cdc68e6139a010c06e8fc96c7 reported by youling257.
Hi,Larry Finger my userspace scan wifi signal request NL80211_STA_INFO_TX_FAILED support, so i make a patch, my userspace scan wifi signal seem OK. Subject: [PATCH] add NL80211_STA_INFO_TX_FAILED support --- drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c b/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c index af2234798fa8..637684b93828 100644 --- a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c +++ b/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c @@ -1278,6 +1278,8 @@ static int cfg80211_rtw_get_station(struct wiphy *wiphy, sinfo->filled |= BIT_ULL(NL80211_STA_INFO_TX_PACKETS); sinfo->tx_packets = psta->sta_stats.tx_pkts; + sinfo->filled |= BIT_ULL(NL80211_STA_INFO_TX_FAILED); + } /* for Ad-Hoc/AP mode */ -- 2.19.1 if you think my patch right, can you add detail and subject it? btw,only these has NL80211_STA_INFO_TX_FAILED https://elixir.bootlin.com/linux/latest/ident/NL80211_STA_INFO_TX_FAILED
(In reply to youling257 from comment #18) > Hi,Larry Finger > my userspace scan wifi signal request NL80211_STA_INFO_TX_FAILED support, > so i make a patch, my userspace scan wifi signal seem OK. > > Subject: [PATCH] add NL80211_STA_INFO_TX_FAILED support > > --- > drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c | 2 ++ > 1 file changed, 2 insertions(+) > > diff --git a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c > b/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c > index af2234798fa8..637684b93828 100644 > --- a/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c > +++ b/drivers/staging/rtl8723bs/os_dep/ioctl_cfg80211.c > @@ -1278,6 +1278,8 @@ static int cfg80211_rtw_get_station(struct wiphy > *wiphy, > sinfo->filled |= BIT_ULL(NL80211_STA_INFO_TX_PACKETS); > sinfo->tx_packets = psta->sta_stats.tx_pkts; > > + sinfo->filled |= BIT_ULL(NL80211_STA_INFO_TX_FAILED); > + > } > > /* for Ad-Hoc/AP mode */ > -- > 2.19.1 > > if you think my patch right, can you add detail and subject it? > > btw,only these has NL80211_STA_INFO_TX_FAILED > https://elixir.bootlin.com/linux/latest/ident/NL80211_STA_INFO_TX_FAILED How do I reproduce this? I'm using Debian-9 and my userspace wifi signal seem fine even after adding that "single" character patch to 4.19.1. Do I need this patch as well?
my userspace is Androidx86 7.1 and 8.1, run Android on Linux mainline kernel. comment 10 and 11, i tested on Androidx86 7.1, not need wificond. but Android 8 wificond, wificond: failed to get nl80211_sta_info_tx_failed WificondControl: Invalid signal poll result from wificond
this is Android wificond problem, impossible every wifi driver all support "NL80211_STA_INFO_TX_FAILED".
Ok, got it. By the way, your previous fix works fine, I now have the normal wifi signal.
Yes, I will prepare a patch. Do I understand correctly that the fix is needed for Android 8? Does it cause any problems if this change is used with 7.1? What kernel is used with Android 8?
(In reply to Larry Finger from comment #23) > Yes, I will prepare a patch. > > Do I understand correctly that the fix is needed for Android 8? Does it > cause any problems if this change is used with 7.1? What kernel is used with > Android 8? Android 8 need NL80211_STA_INFO_TX_FAILED, no any effect on Android 7, google Android 8 used 4.9 lts, google Android 9 used 4.14 lts. Androidx86 project used 4.18/4.19 kernel.
The fix for Android is queued for inclusion.