| 15:16 | topi` | jnettlet: I get this quite regularly when doing "modprobe -r brcmfmac": |
| 15:16 | topi` | [702948.477009] WARNING: driver brcmfmac_sdio did not remove its interrupt handler! |
| 15:17 | topi` | Is this something to be worried about, or just ignore? |
| 15:17 | topi` | I mean, does it grab a handle of a particular IRQ which cannot then be reused? |
| 15:17 | jnettlet | topi`, just ignore i. I believe there is an upstream patch to fix it now |
| 15:17 | jnettlet | it is an ordering problem |
| 15:18 | topi` | I've found that the brcmfmac sometimes stops functioning and then I need to rmmod it and modprobe it again |
| 15:18 | topi` | [703073.223402] brcmfmac: brcmf_c_preinit_dcmds: Firmware version = wl0: Jan 23 2013 17:47:32 version 5.90.195.114 FWID 01-f9e7e464 |
| 15:19 | topi` | seems like a slightly old firmware? |
| 15:19 | jnettlet | it is an old card |
| 15:19 | topi` | BTW, we're probably going to get more orders for HB Gate |
| 15:20 | topi` | it took them a *long* time to realize that yes, there are good *and* cheap solutions :) |
| 15:20 | jnettlet | topi`, great. FYI that we will be revving our SOM soon and switching to a TI WiLINK 8 wifi/bt chip |
| 15:20 | topi` | or then our salesmen are just inept |
| 15:20 | topi` | what do you think of the WiLINK? We had a sampler of those while I was working at Nokia... |
| 15:20 | topi` | honestly I remember nothing about it |
| 15:21 | jnettlet | topi`, much better than the broadcom, uses a tiny bit more power, but much much better reception and throughput |
| 15:21 | jnettlet | we can max out the chip at ~92Mbps |
| 15:22 | topi` | I wonder why the 4330 was such a widely used chip... I think even iPads used it |
| 15:22 | jnettlet | we will also support 1x1 and 2x2 antenna configurations |
| 15:22 | topi` | good |
| 15:22 | topi` | but the metal enclosure needs a drill job ;) |
| 15:22 | jnettlet | yeah the iphone 4s and ipads used it. |
| 15:22 | jnettlet | back when iphones used to have horrible wifi problems. |
| 15:22 | topi` | well, the apple products are known to have reception problems ;) |
| 15:23 | jnettlet | topi`, all the hardware will be modified to support the new configurations |
| 15:23 | topi` | OK |
| 15:23 | topi` | let's speculate that we have a customer who needs a 2x2 antenna configuration.. what time line are we looking at? |
| 15:26 | jnettlet | good question. Let me check quick. |
| 15:27 | topi` | and we are suckers for antenna holes, because all our products base on cellular modems... |
| 15:28 | jnettlet | topi`, I have discussed adding plastic panels to the top of the case where the smooth portions are. Does that help or do you do all pole antennas? |
| 15:29 | topi` | usually all pole antennas. But if it's inside a building etc. then WiFi will work for a short range even without |
| 15:32 | topi` | darn, iw wlan0 scan results in nothing. I'm not sure if it's because the AP is out of range (the antenna on HB Gate is a bit small) or if it's because the brcmfmac driver is stuck |
| 15:33 | topi` | another machine nearby reports the signal as -75.00 dBm so it is a bit on the weak side |
| 15:33 | jnettlet | topi`, try it again. Sometimes the brcm4330 firmware can be slow |
| 15:33 | topi` | I guess I should buy a 4G router with proper antennas at home |
| 15:35 | topi` | I need to make an image and possibly make it smaller than the 8 GB card I'm hosting it on |
| 15:35 | topi` | any suggestions on how to resize the rootfs smaller? :) |
| 15:36 | jnettlet | are you using an ext4 filesystem? |
| 15:40 | topi` | yes |
| 15:41 | topi` | it's the SR image downloaded from sr-build.xyz or whatever was the address |
| 15:41 | topi` | well, it's "tailored"... :) |
| 15:42 | jnettlet | sure, so you can resize2fs the partition down to the size you want, then fdisk the partition table deleting the existing partition and recreate it so it just fits the size of the filesystem you shrunk. |
| 15:42 | topi` | I forgot about resize2fs |
| 15:42 | jnettlet | then when you dd it only dd the number of blocks you have filled |
| 15:43 | topi` | but SD images are always tricky to create.. mainly because SD cards themselves are never the same amount of sectors |
| 15:43 | topi` | then you end up with a GPT table that has a corrupt alternate table |
| 15:43 | topi` | until you fix it with "gdisk"... |
| 15:44 | topi` | oh, this seems to be MBR only |
| 15:44 | jnettlet | yeah GPT and many ARM32 chips do not work well together |
| 15:44 | jnettlet | yeah MBR behaves better |
| 15:44 | topi` | those android bootloaders seem to use GPT... |
| 15:45 | topi` | btw, did you see this: http://liliputing.com/2016/05/nxpfreescales-64-bit-mx-8-chips-finally-coming.html |
| 15:46 | topi` | then, it's another matter whether or not continue using i.MX series chips, wrt the amount of bugs they have in silicon |
| 15:47 | topi` | "NXP is positioning the i.MX 8 chip family as a solution for connected vehicles and industrial or medical products. " |
| 15:47 | topi` | sounds like our area of expertise |
| 15:47 | rabeeh | every chip has it's forte |
| 15:48 | topi` | has there been any news about when the i.MX6 "plus" chips are going to be available? |
| 15:49 | topi` | I want to see the weak ram and/or bus performance fixed :) |
| 15:50 | jnettlet | topi`, they don't really fix it. They have kind of worked around it for the GPU and IPU but that is it |
| 15:50 | jnettlet | very disappointing actually |
| 15:50 | topi` | did you benchmark it yourself? |
| 15:50 | topi` | maybe they bumped the revision of the cortex-a9 core at least? |
| 15:50 | jnettlet | yes |
| 15:51 | topi` | the a9r4 should have better prefetching than previous revisions |
| 15:51 | jnettlet | slightly better, but at an increased power draw and thermal footprint |
| 15:52 | topi` | that sounds odd :/ |
| 15:53 | jnettlet | the SOC is substantially larger than the iMX6 |
| 16:09 | topi` | BTW do you know if the "rngd" package from debian is able to use the rng generator inside i.MX6 to provide entropy? |
| 16:09 | topi` | we have noticed that sometimes "hostapd" complains about lack of entropy |
| 16:09 | topi` | I meant "rng-tools" package, not "rngd" |
| 16:10 | jnettlet | topi`, yes it is supported |
| 16:11 | jnettlet | just install the package and it all starts working without any additonal configuration |
| 16:14 | topi` | great |
| 16:14 | topi` | yeah I reckon it is a kernel feature that the user space daemon then just uses |
| 16:15 | topi` | what's the best way to produce audio from mplayer? ALSA or pulseaudio? |
| 16:15 | topi` | I have some negative experiences of pulseaudio but then again, it was many years back |
| 16:16 | topi` | and maybe gst-launch would be a better tool on i.MX6 than the venerable mplayer? |