10:08 | rabeeh | Ke: no idea |
10:08 | rabeeh | i can ping them again |
10:09 | Ke | =o) |
10:09 | Ke | I actually got USB wifi AP working now very well |
10:10 | rabeeh | you are looking for AP? |
10:10 | rabeeh | :) |
10:10 | rabeeh | you can't get high throughput on USB; |
10:10 | Ke | I am planning basically to put everything on one device |
10:10 | rabeeh | we usually use qca for a serious AP (external signal amplifier) |
10:10 | Ke | you can't get high throughput on wifi anyway |
10:11 | Ke | well mostly |
10:11 | rabee | 10:11 * rabeeh can get 1.3Gbps on 4x4 wave2 |
10:11 | rabeeh | we have that on Armada 38x clearfog |
10:11 | Ke | I bet you tied the antennas together |
10:11 | rabeeh | but that's really extremely good line of sight and good conditions |
10:11 | Ke | yes |
10:12 | Ke | my USB wifi actually gave better throughput than the old ath9k_pci |
10:13 | rabeeh | which one is that? |
10:13 | rabeeh | which dongle are you using? |
10:14 | Ke | the pci device is the one recommended by pc-engines and USB one is https://www.amazon.de/dp/B00LLIOT34/ |
10:14 | Ke | ralink rt2800 |
10:14 | Ke | well driver anyway, chip is 57xx IIRc |
10:15 | rabeeh | did it fit with the terminal micro usb cable? |
10:17 | Ke | I don't follow you now, there is no micro USB anywhere |
10:25 | Dr_Herbalist | Good morning people. |
10:26 | Dr_Herbalist | Yokto - I ran "nice bitbake core-image-x11" - I'm getting this: "WARNING: The license listed Firmware-bcm4330 was not in the licenses collected for linux-firmware" |
10:26 | Dr_Herbalist | Any ideas? |
10:27 | Ke | does the warning relate to some actual problem you are facing |
10:27 | Dr_Herbalist | Yeah, I have no wifi drivers on that build |
10:28 | Dr_Herbalist | Well, I think that's the cause. |
10:28 | Ke | yes, it's plausible |
10:28 | Ke | you have no wifi drivers or they fail to work due to missing fw? |
10:33 | Dr_Herbalist | good question |
10:33 | Dr_Herbalist | ill check |
10:47 | Dr_Herbalist | Yep, no driver present |
10:48 | Dr_Herbalist | firmware* |
10:48 | Ke | so driver yes, firmware no |
10:48 | Ke | ? |
10:59 | Dr_Herbalist | I don't know. I'm a noob. I ran lsmod and its not in there, i tried find /sys and tried to grep anything bcm related but cant find anything. |
11:00 | Ke | find /lib/modules/ -iname 'bcm*.ko' |
11:01 | Dr_Herbalist | oh |
11:01 | Dr_Herbalist | thanks |
11:01 | Dr_Herbalist | so i got... |
11:01 | Dr_Herbalist | 5 lines. |
11:01 | Dr_Herbalist | let me type them |
11:01 | Dr_Herbalist | one second |
11:03 | Dr_Herbalist | /drivers/media/dvb-frontends/bcm3510.ko - /drivers/bluetooth/bcm203x.ko - /drivers/bcma/bcma.ko - /drivers/input/mouse/bcm5974.ko - /drivers/net/phy/ncm87xx.ko |
11:04 | Dr_Herbalist | nothing realted to bcm4330 though |
11:04 | Dr_Herbalist | so looks as though they didnt compile? |
11:05 | Ke | find /lib/modules/ -iname 'brcm*.ko' |
11:06 | Ke | apparently broadcom uses the 'r' there |
11:06 | Ke | there was also b43 |
11:07 | Ke | ls /lib/modules/4.9.0-2-amd64/kernel/drivers/net/wireless/broadcom/ |
11:08 | Ke | where 4.9 is your driver version |
11:16 | Dr_Herbalist | for find - net/wireless/brcm80211/brcmutil/brcmutil.ko - /net/wireless/brcm80211/brcsmac/brcmsmac.ko - /net/wireless/brcm80211/brcfmac/brcmfmac.ko |
11:17 | Dr_Herbalist | i have /b43/b43.ko |
11:34 | jnettlet | Dr_Herbalist, you need to add LICENSE = "Proprietary" to your local.conf |
11:35 | Dr_Herbalist | ahh... |
11:35 | Dr_Herbalist | I added LICENSE = "commercial" |
11:36 | Dr_Herbalist | Id also set it as LICENSE_FLAGS_WHITELIST = "commercial" |
11:37 | Dr_Herbalist | thanks jon. |
11:50 | Dr_Herbalist | I added that line, still the same WARNING though. |
11:50 | jnettlet | Dr_Herbalist, which version of yocto? |
11:51 | Dr_Herbalist | latest one from the git, pulled it down yesterday |
11:51 | Dr_Herbalist | same with the solid run git |
11:52 | jnettlet | yes but which branch? fido, jethro, morty? |
11:52 | Dr_Herbalist | oh |
11:52 | jnettlet | I would recommend using jethro. morty still has a couple of issues I need to iron out. |
11:53 | Dr_Herbalist | morty rings a bell but ill check |
11:53 | Dr_Herbalist | oh it was fido |
11:53 | Dr_Herbalist | I must be thinking of rick and morty |
11:53 | Dr_Herbalist | lmao |
11:53 | Dr_Herbalist | But yeah, defo fido |
11:55 | jnettlet | okay doing a test build. |
11:56 | Dr_Herbalist | Thanks Jon, I really appreciate that |
11:58 | jnettlet | hmmm. no warning here. |
11:59 | jnettlet | this is my local.conf https://pastebin.com/zeQTQpri |
11:59 | jnettlet | and this for machine. MACHINE ?= "solidrun-imx6" |
12:01 | Dr_Herbalist | oh |
12:01 | Dr_Herbalist | i didnt append any of those lines |
12:01 | Dr_Herbalist | that is probably the issue then |
13:14 | jnettlet | Dr_Herbalist, did that fix things up for you? |
14:16 | Dr_Herbalist | hi jnettlet, Just compiling now, fingers crossed |
14:56 | Dr_Herbalist | still executing runqueue |
14:56 | Dr_Herbalist | It must be doing something different at least. |
15:01 | Dr_Herbalist | hmm |
15:01 | Dr_Herbalist | it spat out the same warnings, as well as a few others |
15:01 | Dr_Herbalist | but it ran through a couple hundred extra packages |
15:01 | Dr_Herbalist | so perhaps it worked |
15:03 | jnettlet | well if it isn't working let me know. |
15:04 | Dr_Herbalist | thanks jnettlet, you da real MVP |
15:04 | jnettlet | glad to get that sorted out for you..hopefully |
15:17 | topi` | damn scp is slow over gigabit ethernet... the imx6 seems to cap transfers to around 12MB/s |
15:17 | topi` | we need to compile openssh to use the kernel crypto facilities... kernel crypto should have support for the i.mx6 crypto engine |
15:34 | topi` | compiled openssl with HAVE_CRYPTODEV but it wants crypto/cryptodev.h ... |
15:34 | topi` | hmm |
15:38 | jnettlet | yes, you need to download and point it to cryptodev.h |
15:38 | jnettlet | although you will get better performance just enabling NEON kernel crypto support |
15:38 | topi` | really? |
15:39 | topi` | hard to imagine that the debian build would not use such optimizations |
15:39 | topi` | but i'll check |
15:39 | jnettlet | our kernel should, but I thought you were building your own |
15:40 | topi` | yeah, there's NEON implementation inthe kernel crypto api |
15:40 | topi` | I thought you could use NEON without the kernel api |
15:42 | jnettlet | well you still need cryptodev. sorry to make that sound like they should be exclusionary |
15:43 | topi` | ah, BSAES_ASM is bitsliced aes using neon |
15:43 | topi` | but I'm not sure bitsliced aes can be used with openssh |
15:43 | topi` | I wonder if there's an easy way to disable encryption from ssh |
15:43 | topi` | probably not |
15:44 | topi` | I guess I could use a pipe ending with "nc" to do stuff without encryption, but then I'd have to set up some listener on the remote side |
15:45 | jnettlet | you should probably run ssh in verbose mode and see which algorithm is being tried. I know I had to tweak things to prefer AES algorithms |
15:50 | topi` | yeah. I specified -caes128-ctr for ssh and now it is slightly faster |
15:51 | topi` | I wish I could temporarily overclock the cortex-A9's to 1.2 ghz ;) |
15:51 | topi` | I guess the imx6 cores were designed for 1.2 ghz |
15:58 | jnettlet | topi`, you can try my optimized glibc if you want. I can get transfers from 17MB - 19MB/s |