| 08:40 | bencoh | mhilt: yes I have used parallel csi ov5640 with imx6/7, but not quad |
| 08:41 | bencoh | I'm not sure the driver for the csi controler present in dual/quad has been ported to subdev/ . If not you'll have to use the one in capture/ |
| 09:09 | jnettlet | bencoh, it has not. |
| 09:10 | jnettlet | but it does look like it should work fine in parallel mode |
| 10:08 | topi` | any reason why my custom compiled kernel works on the old skool HB Pro, but fails to mount root at all on a HB Gate? |
| 10:08 | topi` | Please append a correct "root=" option; here are the available partitions: |
| 10:09 | topi` | and then it lists ram0 .. ram15 and nothing else |
| 10:09 | topi` | where oh where is my mmcblk0 ??? |
| 10:12 | topi` | I think the imx6dl-hummingboard2.dtb is correct and in the right place, but I wonder how UBOOT detects which .dtb it needs to load? |
| 10:28 | jnettlet | topi`, you are using an upstream kernel? |
| 10:28 | jnettlet | if so then they changed the device enumeration |
| 10:54 | topi` | yeah, the 4.11 |
| 10:54 | topi` | I understand that it's been changed, but what I don't understand is why it only shows ramdisks and nothing else |
| 10:54 | topi` | which makes me suspect that it can't load the right .dtb |
| 10:58 | bencoh | jnettlet: indeed it should |
| 11:26 | jnettlet | topi`, u-boot will tell you which dtb it is loading on the console before launching the kernel |
| 11:27 | topi` | it all happens very quickly :) |
| 11:28 | topi` | I'll tell my coworkers to install a usb-serial cable |
| 11:30 | jnettlet | yes, usb-serial is the only way to go. |
| 11:30 | bencoh | by the way you should be able to check dtb values in /sys/firmware/devicetree, in case you need to |
| 11:31 | topi` | bencoh: we can't get the kernel to mount root, so that's not an option :) |
| 11:31 | topi` | hmm, maybe provide an initramfs with shell... |
| 11:32 | bencoh | ah |
| 11:33 | jnettlet | topi`, which distro are you using? |
| 11:33 | topi` | jnettlet: my colleague claims that the uart (J25) is missing the pins. Is this possible :) |
| 11:34 | topi` | jnettlet: I made an image on top of debian/stretch |
| 11:34 | jnettlet | topi` I can't imagine that. |
| 11:34 | jnettlet | can they take a picture? |
| 11:34 | topi` | I have one HB-Gate at home and it does have the UART pins. |
| 11:34 | topi` | this must be some very early batch of Gates :) |
| 11:34 | jnettlet | we populate all of them since we don't provide jtag |
| 11:35 | jnettlet | anyways there are other uarts. if need be I can hack together a u-boot quick that uses them. |
| 11:35 | topi` | well, you just solder some pins there :) |
| 11:35 | topi` | the easiest way |
| 11:35 | jnettlet | or that |
| 11:36 | jnettlet | or just solder the cable :) |
| 11:36 | jnettlet | there are solutions |
| 11:36 | topi` | the stupid suits in our company told us soldering isn't allowed in premises, since it releases toxic fumes |
| 11:36 | jnettlet | topi`, did you get 3d acceleration working? |
| 11:36 | jnettlet | lol |
| 11:36 | topi` | our corp is ruined by these marketing men :( |
| 11:36 | topi` | no 3d, but I am currently building mesa and libdrm from source |
| 11:37 | topi` | mesa takes a LONG time on the hb :) |
| 11:37 | jnettlet | do you have an ssd? it speeds things up a lot. |
| 11:37 | topi` | yes |
| 11:37 | topi` | compiling on a SDcard is a total no go :) |
| 11:38 | jnettlet | how is the ssd performance with 4.11. last time I checked upstream was 10-20MB/s slower |
| 12:17 | jnettlet | topi`, would it help if we added a mainline kernel and the bits needed use etnaviv to our build server? |
| 12:17 | jnettlet | I have been talking about it for a while. |
| 12:17 | jnettlet | although I would most likely base it on the latest LTS |
| 12:29 | topi` | jnettlet: I'm unsure, because this is a slooow Kingston 32GB SSD |
| 12:29 | topi` | regarding the mainline kernel, it's a good idea to base it on LTS |
| 12:30 | topi` | 4.9 is the LTS |
| 12:30 | mhilt | bencoh & jnettlet: thanks -- I have CONFIG_MXC_CAMERA_OV5640 set, and I see ov5640.c in /capture, but it seems to build from /subdev |
| 12:32 | jnettlet | mhilt...lol. Well that is NXP patches for you. Yes they have used the same Kconfig name for both. |
| 12:33 | jnettle | 12:33 * jnettlet shakes head and laughs to himself |
| 12:33 | jnettlet | oh mhilt. check your .config and make sure SOC_IMX6SX is not selected |
| 12:33 | jnettlet | that should fix the problem |
| 12:33 | mhilt | yeah I was just seeing that ... |
| 12:33 | mhilt | strange this is it isn't selected |
| 12:48 | mhilt | eh, sorry - grep'd the wrong sandbox ... still early here, and I need some coffee! |
| 13:37 | bencoh | mhilt: CONFIG_MXC_CAMERA_OV5640 builds both of them |
| 14:25 | topi` | I noticed I have a 12V 1A power supply in my HB Gate ... I wonder how many +5V amps can the HB's power supply produce out of this 12V? |
| 14:25 | topi` | it might be enough to run one USB 4G modem, but maybe not two... which would explain why the first one stops working properly after I add the second |
| 14:26 | topi` | maybe I'd need at least 1.5A at 12 volts |
| 14:36 | jnettlet | topi`, probably yes. allthough USB can be picky. I would use the USB-OTG and a USB port to help balance things. |
| 14:36 | jnettlet | you can use the top and bottom ports next to the ethernet adapter |
| 14:36 | jnettlet | or the top port, and then any of the others. |
| 15:37 | topi` | jnettlet: we received today a batch of HB Gates (15 pcs), which seem to be dead (the green LED does not light up)... any hints on how to debug? |
| 15:37 | topi` | maybe they are not fused to boot from SD? |
| 15:38 | jnettlet | topi`, does the red light turn on? |
| 15:38 | jnettlet | possibly |
| 15:38 | topi` | the red light turns on |
| 15:38 | jnettlet | okay sounds like they aren't fused |
| 15:38 | topi` | but I was expecting to see the green one, when kernel boots |
| 15:38 | jnettlet | do they have eMMC on board? |
| 15:38 | topi` | I'll dig the exact SoM code to see what it is |
| 15:38 | jnettlet | if so then we don't fuse them |
| 15:38 | topi` | yes, 8 GB eMMC |
| 15:39 | jnettlet | have you built a usb otg recovery cable yet? |
| 15:41 | rabeeh | http://wiki.solid-run.com/doku.php?id=products:imx6:microsom:imx6-fuse-developers#blowing_fuses_to_boot_from_micro_sd |
| 15:41 | rabeeh | read this; |
| 15:41 | rabeeh | download the HERE-1.1 link; very convenient |
| 15:42 | rabeeh | will get you booted quickly; but you need a usb host to host connector (or build one) |
| 15:43 | topi` | yes, my coworkers will try to build such cable |
| 15:44 | topi` | anything else needed than to cross over the D+ and D- signals? |
| 15:45 | topi` | oh, and we want the fuses to boot from eMMC |
| 15:45 | topi` | SD cards are not very robust in our application... |
| 15:45 | topi` | but then we need to do the initial eMMC writing at the same time as eFusing |
| 15:46 | topi` | can we boot a linux kernel w/ initrd "on-the-fly"? |
| 15:49 | topi` | ok, reading the doc makes it clear that you can upload zImage via usb-otg |
| 15:55 | topi` | hmm, it seems we don't get MAC addresses either? we need MACs for the eth0 and especially the wlan0 |
| 15:55 | topi` | eth0 we can give a random mac, it's not used in "production" |
| 16:06 | jnettlet | topi`, you can specify your own, or write to sales and they will allocate you a group from our block |
| 16:08 | topi` | OK |
| 16:18 | topi` | jnettlet: did you integrate your usb-otg regulator changes to the 3.14.x kernel at some point? or do I need to apply the patch you sent me :) |
| 16:19 | topi` | there's a dead LTE modem sitting on my Gate now, and it'd be convenient to actually take the VBUS from it instead of rebooting the whole device ... |
| 16:20 | jnettlet | topi`, they are merged |
| 16:21 | topi` | cool. so I'll just compile the newest 3.14.79+ |
| 16:24 | jnettlet | topi`, those patches should apply to upstream without much of an issue |
| 16:26 | topi` | yes, I know |
| 16:26 | topi` | but since I have many boards, I often just take the SR provided image and write it to a new SD card in order to boot a new board... |
| 16:27 | topi` | btw, changing the power supply from 1A @12V to 2A @9V solved the problem with the LTE modem |
| 16:27 | topi` | now it connects |
| 17:07 | topi` | jnettlet: after downloading a few megs, the LTE modem stops dead and I see this in dmesg: |
| 17:07 | topi` | hub 2-1.1:1.0: cannot reset port 2 (err = -71) |
| 17:07 | topi` | cannot enable port 2. Maybe the usb cable is bad? |
| 17:08 | topi` | the LTE is hooked to the upper usb port of those two next to the mikroBUS ... so, opposite of the mains power etc |
| 17:08 | topi` | is this usb_h1 or usb_otg? |
| 17:09 | topi` | oops, lsusb causes "task kworker blocked for more than 120 seconds" |
| 17:11 | topi` | I wonder if there's a way to map these regulator names (usb_otg, usb_h1) to those usb 2-1, 1-1 etc ids? |
| 17:18 | jnettlet | osb_otg |
| 17:18 | jnettlet | Ahh... usb_otg |
| 17:38 | topi` | are the usb signals going to the mPCIe slot paired to the usb_otg controller or usb_h1? |