05:17 <jernej> fcas: is this your board? :) https://machinehum.medium.com/im-putting-a-wifi-router-into-a-wall-charger-part-1-882df714bbf3

05:18 <jernej> sorry, I mean machinehum ^

10:41 <warpme_> What xr819 firmware are you using? Mine is quite old downloaded from armbian firmware repo.

10:41 <warpme_> jernej: i have issue with getting cw1200 on xr819 working ok in normal usage. All is perfect with connecting to AP and transfering some data at begining (boot of my appliance + download some config files) but this is all. after that wifi connections seems to silently die. I can't connect via telnet to examine dmesg etc. Is cw1200 on xr819 working stable for you? I mean i.e you can you transfer 1GB of data without issue?.

10:43 <jernej> warpme_: I didn't test it much

10:43 <jernej> I have two FWs, one from karabek repo and one dumped from my TV box Android image (~2 years newer)

10:44 <warpme_> sure. just trying to give you some feedback for v.nice work on cw1200.

10:44 <jernej> with newer there is one complain when debug info is enabled

10:44 <jernej> maybe you can try it?

10:44 <warpme_> is android fw. available somwhere?

10:44 <jernej> I'll upload it

10:46 <jernej> warpme_: http://jernej.libreelec.tv/xr819-fw/

10:46 <warpme_> qll. thx. let me try :-)

10:47 <jernej> currently I'm trying to convince wfx driver to work with xr819

10:47 <jernej> it's much better maintained

10:47 <jernej> but for some reason bootloader upload doesn't work...

10:48 <jernej> it will take some time

11:07 <jernej> warpme_: if you put #define DEBUG 1 at the very top of drivers/net/wireless/st/cw1200/wsm.c it will produce log, which could be helpful in fixing issues

11:07 <warpme_> perfect. let me try this!

11:41 <warpme_> jernej: just done some tests: 1\new fw not helps. 2\issue with died wifi comm. is result of driver trap. here is log: https://pastebin.com/VGBcbWgd

11:42 <warpme_> trap happens after some usage. initially all is ok but after transmitting some data - there is oops....

11:42 <jernej> ok, thanks, I'll take a look

11:42 <jernej> warpme_: is there an easy way to trigger it?

11:45 <warpme_> for me it was very easy: i configure my appliance to boot to term. after boot i open remote telnet session; in remote telnet i asked for dmesg and got oops; in local term i copied dmesg to sd card to provide it to you.

11:46 <warpme_> but i'm pretty sure any play with wget or netperf will be enough

11:49 <jernej> interesting, I did "curl google.com" via serial console and it worked ok

11:49 <jernej> maybe it just needs some more data

11:50 <warpme_> or i.e. it is mtu like issue?

11:53 <warpme_> my appliance uses wget to download config data and those downloads are ok. but starting telnet session and i.e. asking for simple thing like listing hangs. interesting is that i have _exactly the same_ issue on one of s912 boxes with qca9377 chip.

12:05 <jernej> I really don't know

12:13 <jernej> arnd: wfx driver seems to use a bit different FW API - it doesn't work out of the box with XR819

12:35 <jernej> fun fact - each of these ST-Ericsson WIFI descendants have very similar FW API, except that each vendor thought to tweak it a bit to improve (?) certain things

12:36 <jernej> so there will be some "if (hw == xr819)" things...

12:37 <jernej> warpme_; I see API difference in *_tx_confirm_cb() functions

12:42 <warpme_> ah good - this can be good anchor!

12:43 <jernej> I'm just not sure how relevant it is

12:54 <jernej> warpme_: try with changing 0xff to 0xf in cw1200_queue_get_queue_id() in drivers/net/wireless/st/cw1200/queue.h

13:10 <jernej> warpme_: I updated https://github.com/jernejsk/linux-1/commits/h616-test with additional changes which should help with xr819

13:10 <jernej> please test

13:12 <warpme_> jernej: will do asap. (in the meantime I launched mesa/mythtv rebuild so it will take 20min or so)

13:12 <jernej> no hurry :)

13:42 <jernej> warpme_: I got few hiccups when downloading ~400 MiB file, but md5 checksum matches

13:42 <jernej> and speed was pretty decent, 1.5 - 3 MiB/s from remote server

13:43 <jernej> (currently I'm not very near to wifi router)

16:21 <jernej> warpme_: any issues with xr819?

17:20 <warpme_> jernej: i can confirm. current code works very well! So far i played 10min of hd video but all seems to be good. btw: do you have plan to go with cw1200 to mainline or rather your plan is to go with wfx driver?

19:46 <jernej> warpme_: good to hear

19:47 <jernej> currently plan is with cw1200 since FW API is much closer, basically the same

19:47 <jernej> but we'll see

19:48 <jernej> warpme_: do you have any feeling if cw1200 driver is any better or worse to karabek driver?

19:51 <warpme_> great. only thing i think is worth to somehow address is mac addr. I can't get it constant even with alias. in this context karabek driver has nice feature of declaring mac in dt.

19:52 <warpme_> another thing is cw is in-tree - which is HUGE benefit!

19:57 <warpme_> btw: on cw driver - after 1h of hd playback dmesg has:

19:57 <warpme_> https://www.irccloud.com/pastebin/BG5ZBwHq/

19:58 <warpme_> no issues with playback. just above dmesg entries

19:58 <jernej> yeah, I got that too, but as you said, no noticable issue

19:59 <jernej> I guess it's ok for initial bring up :)

19:59 <jernej> but I'll try to track it down

19:59 <jernej> I'll take a look for mac too

20:00 <jernej> note that this driver predates OF in Linux

20:04 <warpme_> oh for sure it is perfect candidate for initial bringup. I think it is worth to consider to adjust fw to karabek path & filenames. This will make switch from karabek to cw seamless

20:05 <jernej> we'll see, ideally, company which produces wifi chip submits FW to linux-firmware repo

20:05 <jernej> I'm not sure xradio will have any incentive to do that

20:05 <jernej> (that also implies proper names)

20:08 <warpme_> exactly. last 5y shows chip producer has zero willingness to have this chip supported by mainline linux. by this i will absolutely ignore any aspect of them....

21:56 <apritzel> warpme_: the MAC address "feature" is a generic DT feature: U-Boot will put the proper MAC address into "local-mac-addr" property of the device DT node

21:56 <apritzel> and a call to of_get_mac_address() from within the MAC driver will parse this

21:57 <apritzel> as jernej mentioned, the cw driver is predating DT, so it's missing all those generic DT features

22:00 <jernej> yeah, this will be easy to fix

22:48 <apritzel> did anyone try SPI booting on the OrangePi Zero 2 (H616)?

22:49 <apritzel> I fixed the U-Boot part, but the BROM won't honour my eGON

22:50 <apritzel> I hacked sunxi-fel to fake SPI boot, and this part works (the SPL code loads TF-A and U-Boot), but the BROM drops me to FEL

22:52 <apritzel> smaeul: do you already have a start address in the H616 BROM where SPI is handled?

23:42 <smaeul> apritzel: I'm travelling atm, so I don't my usual tools. the SPI0 address is at 0x6a74, so I'd start looking at references to there (tbh if you're using ghidra, this should be obvious).

23:46 <apritzel> smaeul: yeah, found that place as well, will have a look in there

23:47 <apritzel> smaeul: enjoy the fireworks ;-)