| freemangordon1 | Wizzup: yeah, no chanlog there | 08:02 |
|---|---|---|
| stano_ | droid4: messages wlcore: Association completed. | 11:30 |
| stano_ | omap-mcbsp 40124000.mcbsp: CLKS: could not clk-get() prcm_fck | 11:30 |
| stano_ | oh it eventually connected, nvm | 11:31 |
| stano_ | greets - (droid4) is the battery-charging indicator always showing charging for anyone else? | 13:21 |
| Wizzup | looks like the wl1251 did get some fixes since our kernel, but none seem to relate | 13:32 |
| stano_ | now it stopped always showing charging 0_o | 13:36 |
| stano_ | maybe because i'm in a long compile | 13:37 |
| Wizzup | Pali: ping | 14:07 |
| Pali | Wizzup: pong | 14:07 |
| Wizzup | looking at the scan problem in wl1251, where on nl80211 the wpa_supplicant scan requests get EINVAL, PaulFertser from #linux-wireless suggested to add wl->hw->wiphy->max_scan_ie_len to wl1251 | 14:08 |
| Wizzup | I am looking at wlcore and it does set that, but it has a constant like htis: #define WL1271_CMD_TEMPL_MAX_SIZE 512 | 14:08 |
| Wizzup | any idea where I could find the max command template length for wl1251? | 14:08 |
| Pali | I have no idea what this template length is | 14:09 |
| Wizzup | (looking at wlcore for an other driver that seems very similar) | 14:09 |
| Wizzup | me neither :) | 14:09 |
| Pali | anyway, in past when I debugged wl1251 kernel driver I was looking into TI SDK driver for wl1251, so maybe it would be again | 14:10 |
| Wizzup | ok | 14:10 |
| Pali | wl1251 is marked as "TI WiLink 4.0" | 14:11 |
| Pali | somewhere on internet was TI driver for this WiLink 4 | 14:11 |
| Pali | I will try to find it | 14:12 |
| Wizzup | btw, this change makes wl80211 scanning work properly: | 14:19 |
| Wizzup | it's probably not a correct fix: | 14:19 |
| Wizzup | wl->hw->wiphy->max_scan_ssids = 1; | 14:19 |
| Wizzup | + wl->hw->wiphy->max_scan_ie_len = 512; | 14:19 |
| Wizzup | wl->hw->wiphy->bands[NL80211_BAND_2GHZ] = &wl1251_band_2ghz; | 14:19 |
| Wizzup | heh I wrote wl80211 , that should be nl80211 | 14:20 |
| Wizzup | freemangordon: I think that the hidden ap scan via connui doesn't work because of some perms problems | 14:33 |
| Wizzup | string "Rejected send message, 2 matched rules; type="method_call", sender=":1.28" (uid=1000 pid=2714 comm="/usr/bin/osso-connectivity-ui-conndlgs ") interface="fi.w1.wpa_supplicant1.Interface" member="Scan" error name="(unset)" requested_reply="0" destination="fi.w1.wpa_supplicant1" (uid=0 pid=2791 comm="/sbin/wpa_supplicant -s -P /run/wpa_supplicant.wla")" | 14:33 |
| Wizzup | freemangordon: after I added user to the policy, it works ! | 14:35 |
| Wizzup | freemangordon: did you change /etc/dbus-1/system.d/wpa_supplicant.conf at all for conndlgs ? | 14:42 |
| Wizzup | parazyd: should user be group netdev? | 14:42 |
| Wizzup | I think it should be | 14:43 |
| Wizzup | debian says: | 14:43 |
| Wizzup | netdev: Members of this group can manage network interfaces through the network manager and wicd. | 14:43 |
| Wizzup | netdev would allow user to issue dbus calls to wpa_supplicant | 14:45 |
| parazyd | I thought wpa_supplicant uses wheel for control interface | 14:47 |
| Pali | Wizzup: here is "original" TI SDK driver for wl1251: https://android.googlesource.com/platform/system/wlan/ti/+/refs/heads/master | 14:52 |
| Pali | header files in this driver contains documentation of RPC commands supported by wl1251 firmware | 14:53 |
| Pali | and IIRC it is the only public documentation for wl1251 firmware | 14:53 |
| Wizzup | parazyd: this is about dbus | 14:54 |
| Wizzup | parazyd: and it is visible in /etc/dbus-1/system.d/wpa_supplicant.conf | 14:54 |
| Wizzup | Pali: ty | 14:54 |
| Pali | WiLink 4 = wl1251; WiLink 6 = wl271 | 14:54 |
| Wizzup | Pali: I think I've actually gotten a "proper" fix now, but will compare with that driver :) | 14:54 |
| Pali | 6 = wl1271 | 14:55 |
| parazyd | Wizzup: Then I guess we should | 14:55 |
| Pali | and these firmware shares lot of common commands | 14:55 |
| parazyd | Wizzup: Please make a ticket if you don't want to do it now (it's in hildon-base) | 14:55 |
| Wizzup | parazyd: I don't mind modifying the groups now | 14:55 |
| Wizzup | let me test it | 14:55 |
| parazyd | https://github.com/maemo-leste/hildon-base/blob/master/debian/postinst | 14:56 |
| Wizzup | parazyd: yeah have it open already :) | 14:56 |
| parazyd | Note that it currently doesn't do anything if the user exists | 14:56 |
| parazyd | So you'll need an additional stance | 14:56 |
| Wizzup | hm | 14:56 |
| parazyd | gpasswd -a || true | 14:56 |
| parazyd | Will work | 14:56 |
| Wizzup | is there a reason not to just always go passwd ? | 14:57 |
| parazyd | Not really | 14:57 |
| parazyd | I just didn't want to modify existing user | 14:57 |
| Wizzup | any reason not to do it now? | 14:57 |
| parazyd | No | 14:57 |
| parazyd | Just do it only for _appending_ groups | 14:57 |
| Wizzup | ok | 14:57 |
| Wizzup | yeah, check | 14:57 |
| Wizzup | parazyd: I am not going to add || true | 14:58 |
| Wizzup | since: | 14:58 |
| Wizzup | # gpasswd -a merlijn wheel | 14:58 |
| Wizzup | Adding user merlijn to group wheel | 14:58 |
| Wizzup | # echo $? | 14:58 |
| Wizzup | 0 | 14:58 |
| Wizzup | (I was already in wheel) | 14:58 |
| parazyd | Ah alright | 14:59 |
| parazyd | Wasn't sure that's the case | 14:59 |
| parazyd | (returning 0) | 15:00 |
| Pali | here is that firmware API / RPC documentation file: https://android.googlesource.com/platform/system/wlan/ti/+/refs/heads/master/sta_dk_4_0_4_32/common/src/hal/FirmwareApi/public_infoele.h | 15:00 |
| Pali | seems that their firmware RPC API shares some structures with old TI ACX100/110 wifi chips | 15:03 |
| sicelo | nice info @Pali | 15:04 |
| Pali | Wizzup, freemangordon: maybe you can write some reply to n900 u-boot email? | 15:05 |
| Wizzup | Pali: I am not sure what to reply, I don't think I can do it now or any time soon, but maybe we can find some timeline where we can agree it is doable by one of us | 15:06 |
| Wizzup | parazyd: pushed to hildon-base and building for maemo/beowulf-devel | 15:09 |
| parazyd | Awesome | 15:09 |
| uvos | nice work Wizzup @ wl1251 | 15:13 |
| Wizzup | uvos: it was mostly someone else guiding me, but yeah | 15:14 |
| uvos | to me sitting on the sidelines mabye setting wl->hw->wiphy->max_scan_ie_len to some random value thats ignored later looks a bit wierd | 15:14 |
| uvos | maybe i would be better wlcore manged this case better somehow | 15:15 |
| Wizzup | s/wlcore/wl1251 | 15:15 |
| Wizzup | / | 15:15 |
| uvos | but idk | 15:15 |
| uvos | ok | 15:15 |
| Wizzup | as I understand it, the EINVAL doesn't even come from wl1251, it comes from layer before it | 15:15 |
| uvos | wlcore | 15:15 |
| uvos | no? | 15:15 |
| Wizzup | I don't think so, it's not loaded on n900 | 15:15 |
| Wizzup | I think wl1251 is standalone, whereas on d4 the wlcore and others are loaded | 15:16 |
| Wizzup | iiuc | 15:16 |
| uvos | wl1251 isent a wlcore dirver | 15:16 |
| uvos | ok | 15:16 |
| Wizzup | wlcore *does* set this value | 15:16 |
| uvos | til | 15:16 |
| uvos | ok | 15:16 |
| uvos | Wizzup: so everything should work now? and we are using the command interface of wpa now, not dbus? | 15:21 |
| uvos | if so ill go around testing the current code on the divices i have | 15:22 |
| Wizzup | I need to build fixed kernel and push out some stuff to -devel | 15:22 |
| Wizzup | so maybe wait just a bit | 15:22 |
| uvos | ok when you do ill try xt875/wl1271, n900 wl1251 and d4's wl1285c | 15:23 |
| uvos | particulary i assume no one has tried the new code on wl1271 at all... | 15:23 |
| Wizzup | what 'new code'? | 15:24 |
| uvos | all changes since the start of you working on hidden aps | 15:24 |
| Wizzup | I tested on the d4, not the bionic | 15:24 |
| uvos | right | 15:24 |
| Wizzup | uvos: if you want to test, now on -devel you can test wifi for the devices that are not the n900 | 15:31 |
| Wizzup | well, wait, no | 15:32 |
| Wizzup | uvos: yeah, wait, I need to revert back to libwpa_client once this n900 kernel builds | 15:34 |
| freemangordon | Wizzup: yes, I changed the policy, but IIRC we agreed that 'user' will be netdev member | 17:16 |
| Wizzup | freemangordon: ok, that didn't happen yet, but I did it now | 17:16 |
| Wizzup | freemangordon: on n900 hidden ap stuff works fine now :) | 17:17 |
| freemangordon | great! | 17:17 |
| freemangordon | wext? | 17:17 |
| Wizzup | nl80211 | 17:17 |
| freemangordon | or you fixed the driver | 17:17 |
| freemangordon | nice | 17:17 |
| sicelo | yay | 17:17 |
| stano_ | does this fix losing wifi and all access points disappearing on droid4? | 17:41 |
| stano_ | is there a script to restart wifi driver without rebooting? | 17:41 |
| Wizzup | uvos: I think early next week I'd like to toy a bit with one of the tablets, or the droid 3, see how far I can come, anything you can recommend? | 17:58 |
| Wizzup | I know tony was working on display driver for one of them | 17:58 |
| uvos | Wizzup: the d3 might be easier as the only real problem it should have judging by a breif look at its bord source and teardown images is bootstraping the thing, as it shares the pannel and everything else with the bionic | 18:00 |
| uvos | Wizzup: the tablets (all of them) have a dsi->lvds bridge chip that isent working for unkown reasons, tony was/is working on that. | 18:00 |
| freemangordon | Wizzup: https://github.com/maemo-leste/osso-abook/commit/d524b942918420044348923264229db762e82031 :) | 18:01 |
| Wizzup | freemangordon: amazing, great :) | 18:02 |
| Wizzup | uvos: d3 sounds like fun | 18:04 |
| uvos | right so for d3 i would to look around on xda for the safestrap files and install that, then modify a bionic dts so the memory map ends lower, force cmdline to something and then use mkbootimg to build a boot.img for it and try and boot that | 18:05 |
| uvos | if that fails you have to diff the board src for android 2.3 between targa and solana | 18:06 |
| uvos | since solana never got 4.0 or 4.1 | 18:06 |
| uvos | if it works you can build kexec.ko for the soloana kernel and then clown boot should be a possibility | 18:07 |
| Wizzup | ok | 18:09 |
| parazyd | What's solana? | 18:18 |
| parazyd | (Because https://solana.com/ :p ) | 18:19 |
| uvos | the droid 3's codename | 18:19 |
| parazyd | ah | 18:20 |
| uvos | looks like Hashcode (main mapphone los dev) had solana booting the newest motorola kernel see mapphone_solana_phablet_defconfig in the los kernel and the memory map is just 512 MB starting at 0x80000000 | 18:29 |
| Wizzup | uvos: looks like this mirror is still avail https://mega.nz/folder/dJ4TSbrC#s5zdH-HtDCITKg2xoYS_qQ | 18:58 |
| uvos | Wizzup: any luck with that? | 21:59 |
| Wizzup | uvos: no, didn't do much beyond finding the files yet | 23:22 |
| Wizzup | I might need to root it first | 23:26 |
| Wizzup | there is also https://docs.google.com/document/d/1suBx1DH7jWmHeM0mDerIStodDSgVAeQXxSkaLPlBAAw/edit# | 23:27 |
| stano_ | ppsspp was working. now it not working. :/ http://0x0.st/-VXd.mp4 | 23:45 |
| stano_ | must continue fighting, like hero | 23:46 |
| Wizzup | uvos: latest libicd-network-wpasupplicant is now in -devel | 23:54 |
| uvos | Wizzup: great | 23:59 |
| uvos | bit late to test now but will try on monday | 23:59 |
| uvos | yes you must root first to install safestrap | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!