03:18 <tehchubbz0r> anyone around that could help me with some questions about kernel compiling (not the act of, i've got an issue with a wifi driver that I'm trying to work out)? i'd greatly appreciate any help i can get

07:43 <stdint> tehchubbz0r, what exactly you meet

12:18 <tehchubbz0r> sorry for the late response, I'm compiling source from the rk3188_rk3066_r-box_android4.4.2_sdk branch for an mk809iii with rtl8723bs wifi+bt chipset. I've followed what rockchip documentation I could gather and everything is going well, but I'm having some issues with wifi, the chip powers up, it will associate with an access point and traffic will pass for roughly a 1-2 minutes, and then i start getting sdmmc errors i

12:19 <tehchubbz0r> when this happens, the chip still shows association to the ap, but traffic won't flow and the ap shows the stick isn't associated anymore

12:19 <tehchubbz0r> turning wifi off and back on produces no results, a full reboot of the stick is required to get it to work again

12:21 <tehchubbz0r> i used gpio dump on the stock firmware these sticks ship with and then again on the firmware i built and compared, wifi power, wifi reset, and oob_irq (not sure if used by wifi) are all on the same pins

12:23 <tehchubbz0r> i guess i'm a little lost on where to begin debugging this problem, the fact that the chip powers on and definitely works for a brief period makes me feel that i've got the sdio/mmc drivers set up correctly, and the 8723bs drivers set up correctly, but i don't know if it's a bug/issue in the drivers in the repo, or a misconfiguration on my end

12:36 <tehchubbz0r> i suppose that it could also be a memory leak in the driver, due to the fact that the errors that show up when it stops working are "alloc recvbuf FAIL!" in sd_recv_rxfifo, which through a cursory examination of the source looks like would only be logged if a buffer to hold a received message couldn't be allocated

15:52 <tehchubbz0r> is there a more active time for this channel or is this a pretty typical activity level?

16:09 <mmind00> tehchubbz0r: the issue is most likely that the number of people trying to hack on the old vendor kernels is pretty limited nowadays, so people will most likely only respond to stuff that touches their area of experience

16:11 <tehchubbz0r> mmind00: i can understand that, any advice on tracking down a bug or misconfiguration like this?

16:13 <mmind00> tehchubbz0r: no, sorry ... I only ever looked at the vendor kernel when I started my mainline adventure, but have never tried compiling or running that stuff

16:14 <tehchubbz0r> mmind00: frustrating to know that someone out there was able to compile non-stock firmwares that run on this device but no information on how to get there from the available source

16:26 <tehchubbz0r> perhaps an easier question to get an answer for then, is there a way, short of compiling from source since the rtl8723bs drivers are integrated into the kernel and not a module, to disable the behavior of reducing txpower when other ap's are broadcasting nearby

20:11 <topi`> tehchubbz0r: it's very unfortunate that somebody has decided to compile that as builtin, and not a module

20:11 <topi`> non-" upgradeable"

20:12 <topi`> is there a more up-to-date source for RK mainlining status than linux-rockchip.info wiki?

20:12 <topi`> I'm mostly interested in the RK3368 since I have one Geekbox without a decent OS

20:28 <tehchubbz0r> not that I've seen for the 3188

20:29 <tehchubbz0r> from what i can tell from the code in the repository, it's not available to compile as an add-in module, it's built-in or not at all because the code was modified/written by rockchip, not by realtek

20:30 <tehchubbz0r> it's pretty reliant on the rk29 specific mmc code, so i can see why that decision was made

20:33 <tehchubbz0r> i can't tell exactly where this bug is, but the chip works normally for 60-90 seconds and then stops responding to the system's mmc calls

21:00 <mmind00> topi`: just look into the rk3368.dtsi in your favorite mainline tree to see what is working

21:12 <karlp> any trees floating around yet with 3229 support?