| 01:58 | iio> | Is there a HW mod for cubox to power up a usb device which requires just a tad over 500ma (but less than 1a; I use only 1 usb device, as I understand, 1a in total is reserved for both usb ports) |
| 10:56 | suihkulokki> | 18.06 branch lands to marvell-firmwares - still with power management only for 7040 :-/ |
| 11:17 | topi`> | how does the Marvell wifi driver (wl1800) grab a MAC address for the wlan0 on HB2? |
| 11:19 | topi`> | I mean Texas Instruments, not Marvell |
| 11:22 | jnettlet[m]> | topi`: it is embedded in the wifi firmware. Just make sure not to use the linux-firmware packages they are broken. Grab the firmware files from our repo. |
| 11:22 | jnettlet[m]> | https://github.com/SolidRun/deb-pkg_cuboxi-firmware-wireless |
| 11:25 | jnettlet[m]> | topi`: I assume you are getting the deadbeef mac? |
| 11:26 | jnettlet[m]> | suihkulokki: yeah I am not sure what their software team is doing |
| 11:26 | topi`> | jnettlet[m]: no, I'm getting macs that seem to belong to Texas Instruments mac range |
| 11:26 | topi`> | like 88:c2:55 |
| 11:27 | topi`> | I get: wlcore: wl18xx HW: 183x or 180x, PG 2.2 (ROM 0x11) |
| 11:27 | jnettlet[m]> | that would be correct. |
| 11:27 | topi`> | I also get a weird error |
| 11:28 | topi`> | wlcore: ERROR could not get configuration binary ti-connectivity/wl18xx-conf.bin: -2 |
| 11:28 | topi`> | I assume it will then just use some sane defaults |
| 11:28 | topi`> | kernel 4.14.30 |
| 11:28 | jnettlet[m]> | that is because it is missing :) You need to get the firmware from our repository |
| 11:28 | topi`> | I've manually applied quite a few patchlevels |
| 11:28 | jnettlet[m]> | the config file is chip implementation specific. |
| 11:29 | jnettlet[m]> | so I have generated one that is tuned to our SOM |
| 11:29 | topi`> | great |
| 11:29 | topi`> | so, which is this repo? |
| 11:31 | jnettlet[m]> | I just posted it up above |
| 11:31 | topi`> | sorry :) |
| 11:31 | jnettlet[m]> | np |
| 11:31 | topi`> | do I also need to grab those *sdio.bin files? |
| 11:31 | topi`> | no, brcmfmac is for the older boars |
| 11:32 | suihkulokki> | from the generic distro pov it's infuriating the conf file contents is board specific while the location is static :-/ |
| 11:32 | jnettlet[m]> | those are for the old broadcom based soms |
| 11:33 | topi`> | suihkulokki: a classic screwup from TI |
| 11:33 | jnettlet[m]> | It is the same way for atheros modules |
| 11:33 | topi`> | really? |
| 11:33 | topi`> | I didn't expect that from kvalo |
| 11:33 | jnettlet[m]> | yes, just most people skip the configuration option and use the default internal firmware settings |
| 11:34 | suihkulokki> | I thought atheros had the data in device tree |
| 11:34 | jnettlet[m]> | maybe some of it, but not all of it. |
| 11:35 | Ke> | it should be easy enough to add board name to dtb and and then fetch the conf file based on that |
| 11:35 | jnettlet[m]> | the number of options that can be tweaked on the compex modules is enormous, and most PCIe stuff doesn't end up in the device-tree it just gets auto-loaded |
| 11:35 | Ke> | or even just diff the conf files and apply a binary patch... |
| 11:37 | jnettlet[m]> | well really most distros already have alternatives which should handle this issue |
| 11:38 | Ke> | that is even more horrible way imo |
| 11:39 | jnettlet[m]> | ARM devices are not designed for a generic distribution. They are embedded devices |
| 11:39 | Ke> | I think the problem here is that the configuration blobs are not built by the driver based on dtb |
| 11:39 | jnettlet[m]> | you can't make hyper optimized designs based on a generic software distribution |
| 11:39 | Ke> | as the mdoel would expect |
| 11:40 | Ke> | though obviously dtb might not be suitable for megabytes of data |
| 11:42 | jnettlet[m]> | and many can not be modifiable especially regarding wireless firmware configuration |
| 11:42 | topi`> | what's the best way to build this firmware package? it has the debian/ build files in the hierarchy. Debuild is probably outdated by now? |
| 11:42 | topi`> | I used debuild 10 years ago |
| 11:43 | jnettlet[m]> | debuild should work as far as I know |
| 11:43 | topi`> | it's not found in Stretch |
| 11:43 | topi`> | right now apt-getting debian-builder, but it's downloading a TON of perl packages |
| 11:44 | jnettlet[m]> | well that is debian for ya :) |
| 11:46 | topi`> | oh, I need to install a package called "devscripts" to get debuild |
| 11:50 | topi`> | is it OK to just skip the install of package cuboxi-firmware-wireless-bluetooth-ti_11_armhf.deb |
| 11:50 | topi`> | since we're not using BT? |
| 11:54 | topi`> | wl18xx_driver wl18xx.2.auto: Direct firmware load for ti-connectivity/wl1271-nvs.bin failed with error -2 |
| 11:55 | topi`> | what the heck is that |
| 11:55 | topi`> | I think the chip is of the wl18xx family, not wl12xx |
| 11:59 | Artox> | topi` yes it is okay to skip that bluetooth deb |
| 12:00 | Artox> | and I always run dpkg-buildpackage |
| 18:13 | Guest89407> | Minere bitcoin BTC via CPU https://getcryptotab.com/718967 https://www.youtube.com/watch?v=luzqQN3kL4g&t=166s |