| 09:49 | bencoh | hey there! the clearfog looks pretty cool, but which switch did you choose (for the 6 ethernet ports)? |
| 10:03 | rabeeh | bencoh - Marvell 6176 |
| 10:04 | bencoh | is there a driver upstream? |
| 10:04 | bencoh | (in mainline I mean) |
| 10:04 | jnettlet | bencoh, https://github.com/SolidRun/linux-stable/tree/v4.4-rc6-sr-clearfog |
| 10:04 | bencoh | or is it in specific kernel trees (openwrt I guess) |
| 10:04 | bencoh | thx :) |
| 10:04 | jnettlet | everything will be supported once 4.4 is released |
| 10:05 | jnettlet | but that tree includes rmk's work to add hotplug sfp support |
| 10:05 | bencoh | oh, neat |
| 10:05 | jnettlet | I have two boards connected with a fiber uplink right now |
| 10:06 | bencoh | just out of curiosity, how much trafic/pps did you manage to route (ip forwarding) through CPU (switch -> cpu -> switch)? |
| 10:09 | jnettlet | to the CPU is limited to 1Gbps |
| 10:10 | bencoh | can it actually handle 1gbps? |
| 10:10 | bencoh | (of routed traffic) |
| 10:11 | jnettlet | most likely depends how intense the rules are, and how many layers |
| 10:11 | jnettlet | but it has lots of IO bandwidth. It would come down to CPU |
| 10:12 | bencoh | hmm'kay, it should be better than what we had with the iMX.6 then? |
| 10:13 | jnettlet | oh for throughput it is much much faster |
| 10:13 | jnettlet | you can get a full 550 MBps from an SSD connected to m.2 or mSata |
| 10:14 | bencoh | cool :) |
| 10:16 | bencoh | driver mvneta I guess? |
| 10:57 | heap_ | hi |
| 10:57 | heap_ | what the hell is this... i booted up debian jessie |
| 10:57 | heap_ | and plugged in esata box + ext usb drive |
| 10:57 | heap_ | ext usb drive is not recognized |
| 10:59 | heap_ | ah it needs external power |
| 10:59 | heap_ | ;-( |
| 10:59 | heap_ | 2.5" drive |
| 11:16 | heap_ | looks like lvm doesnt work with kernel that comes with debian jessie install |
| 11:25 | jnettlet | heap_, it should but if it isn't enabled then I will issue a kernel rebuild with that on |
| 11:40 | heap_ | jnettlet: how can i check it? |
| 11:40 | heap_ | jnettlet: any idea what the name line in the config? |
| 11:45 | jnettlet | heap_, you need CONFIG_MD enabled and then device-mapper lvm2 etc |
| 11:54 | heap_ | jnettlet: dont know |
| 11:54 | heap_ | jnettlet: there is no /proc/config.gz |
| 11:55 | heap_ | another missing item :) |
| 11:56 | jnettlet | heap_, our debian builds put the config in /boot |
| 11:56 | jnettlet | no reason to have redundant info |
| 11:56 | heap_ | CONFIG_MD=y |
| 11:57 | heap_ | maybe missing device-mapper lvm2? |
| 11:57 | bencoh | CONFIG_MD doesn't enable lvm |
| 11:57 | bencoh | MD is the raid driver, you need the CONFIG_DM_* |
| 11:57 | bencoh | DM* are the device-mapper drivers |
| 11:58 | bencoh | you might still need _MD for the DM_* infrastructure to work, or at least part of it |
| 11:58 | jnettlet | bencoh, CONFIG_DM isn't available unless you enable CONFIG_MD |
| 11:58 | bencoh | then you need both :) |
| 11:58 | jnettlet | yes |
| 12:01 | heap_ | anyway i run badblocks on disk and during the first pattern check ... compare reported 1 bad block |
| 12:01 | heap_ | (0/0/1) is it because of RAM issue? |
| 12:01 | heap_ | because then i rerun the test and no bad block was reported |
| 13:05 | topi` | jnettlet: I didn't manage to get stuff out of the J35 connector by using ttymxc3 but that might be because of partial electrical connection.. I'd need to solder some pins for that connector to make sure the connection is stable |
| 13:05 | topi` | so, the Mikrobus would be better in that sense - I don't have a soldering station at home :( |
| 13:06 | topi` | jnettlet: the documentation on how to use the different UARTs on HB2 is missing - surely that should be a priority to Solid-run? |
| 13:11 | jnettlet | topi`, how to use them in what way? |
| 13:22 | topi` | jnettlet: we have bought RS485 click boards from Mikroelektronika and I'd need to get them in working order |
| 13:22 | topi` | so far haven't been able to establish working uart comms through the mikrobus uarts |
| 13:52 | heap_ | jnettlet: any idea? |
| 13:53 | jnettlet | heap_, no idea. |
| 14:02 | heap_ | corruption during compare... it looks like os or ram issue to me |
| 14:04 | jnettlet | heap_, this is running our 3.14 kernel? |
| 14:04 | jnettlet | and our u-boot-imx? |
| 14:05 | heap_ | yesyes |
| 14:07 | heap_ | jnettlet: but im going to migrate to xbian 4.1.13 because of btrfs |
| 14:37 | heap_ | so i cant access my lvm anymore |
| 14:37 | heap_ | thats strange |
| 15:43 | heap_ | [ 4079.713799] mxc_hdmi 20e0000.hdmi_video: same edid |
| 15:43 | heap_ | [ 4079.713815] mxc_hdmi 20e0000.hdmi_video: mxc_hdmi_cable_connected_edid reports DVI mode |
| 15:43 | heap_ | any idea about these dmesg messages? |
| 17:23 | heap_ | so what did you guys did to that debian JESSIE? i cant see my LVM inside /dev/mapper |
| 17:24 | heap_ | lvm2 tools are not installed also |
| 17:24 | heap_ | ah |
| 17:24 | jnettlet | nothing. it is stripped down you are probably missing some lvm packages |
| 17:24 | heap_ | yeah |
| 17:24 | heap_ | stripped to bones |
| 17:25 | heap_ | Processing triggers for libc-bin (2.19-18+deb8u1) ... |
| 17:25 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200 is not an ELF file - it has the wrong magic bytes at the start. |
| 17:25 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200.19.0 is not an ELF file - it has the wrong magic bytes at the start. |
| 17:26 | heap_ | it works now, missing lvm2 |
| 17:34 | heap_ | well if i have attached 4 drives by esata and one drive using usb ..i want to boot from that usb drive... but it looks like during the boot that drive has alias once time sda another sde etc... |
| 17:35 | jnettlet | heap_, use the volumes UUID in your fstab |
| 17:35 | heap_ | okey thanks |
| 17:38 | heap_ | any idea what it should be ... |
| 17:38 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200 is not an ELF file - it has the wrong magic bytes at the start. |
| 17:38 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200.19.0 is not an ELF file - it has the wrong magic bytes at the start. |
| 17:44 | jnettlet | heap_, did the file get corrupt? |
| 17:46 | heap_ | how :< |
| 17:46 | heap_ | its fresh install |
| 17:46 | heap_ | or how can i fix it |
| 17:51 | heap_ | there is a bug http://wiki.solid-run.com/doku.php?id=products:imx6:software:development:kernel |
| 17:51 | heap_ | ops |
| 17:51 | heap_ | http://wiki.solid-run.com/doku.php?id=products:imx6:software:development:u-boot&s[]=boot#booting_from_external_device |
| 17:52 | heap_ | mount /dev/sda1 /mnt |
| 17:52 | heap_ | rsync -aH / /mnt |
| 17:52 | heap_ | missing --exclude in rsync |
| 17:54 | heap_ | oh man |
| 17:54 | heap_ | also sys proc etc |
| 17:54 | heap_ | crap |
| 17:54 | jnettlet | And tmp |
| 17:56 | heap_ | im more interested in how to fi those 2 libs |
| 17:56 | jnettlet | Oh I think you can use apt to force reinstall the island that provides them |
| 17:57 | jnettlet | Island? Thanks auto correct |
| 17:57 | heap_ | what provides them? |
| 17:58 | jnettlet | sudo apt-get install --reinstall mypackage |
| 17:58 | jnettlet | dpkg -S /path/to/file |
| 17:58 | heap_ | thankie |
| 18:32 | heap_ | jnettlet: i reinstalled it.. but same issue |
| 18:32 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200.19.0 is not an ELF file - it has the wrong magic bytes at the start. |
| 18:32 | heap_ | /sbin/ldconfig.real: /lib/arm-linux-gnueabihf/libnl-3.so.200 is not an ELF file - it has the wrong magic bytes at the start. |
| 18:33 | heap_ | jnettlet: that debian is really odd |
| 18:33 | heap_ | i even cant install btrfs-tools |
| 18:33 | jnettlet | Perhaps it just reused the same blocks? |
| 18:33 | heap_ | Processing triggers for initramfs-tools (0.120) ... |
| 18:33 | heap_ | update-initramfs: Generating /boot/initrd.img-4.1.13 |
| 18:33 | heap_ | mktemp: failed to create directory via template ‘/var/tmp/mkinitramfs_XXXXXX’: No such file or directory |
| 18:33 | heap_ | update-initramfs: failed for /boot/initrd.img-4.1.13 with 1. |
| 18:33 | heap_ | dpkg: error processing package initramfs-tools (--configure): |
| 18:34 | heap_ | ah |
| 18:35 | heap_ | this is neverending story ;/ |
| 18:36 | heap_ | missing tmp folder inside /var |
| 18:36 | heap_ | strange.. i did rsync... |
| 18:37 | topi` | jnettlet: are all of the UARTs actually muxed in the default dts? |
| 18:37 | topi` | as far as I understand, the kernel will find UARTs just fine, irrespective of the muxings |
| 18:38 | topi` | but if wrong signal is muxed to that pad, nothing moves out of uart |
| 18:39 | vpeter | jnettlet: Did you get email from Christian (chewitt)? |
| 18:41 | jnettlet | topi, yes they are all muxed |
| 18:42 | jnettlet | If not then they show up as gpios generally |
| 18:42 | topi` | OK then I need perhaps an oscilloscope to see what's happening there |
| 18:42 | jnettlet | vpeter, yes I received it. I am trying to sort out with the team how we want to move forward. |
| 18:43 | topi` | btw have the .dtbs been updated for a while? My imx6dl-hummingboard2.dtb is from Oct 28 |
| 18:43 | topi` | and booting this on the imx6solo |
| 18:43 | jnettlet | Have you tried connecting a serial cable to the pins and run agetty on the ttymxc3? |
| 18:45 | jnettlet | That is fine the only difference between there two is a cpu core and memory configuration which uboot takes care of |
| 18:46 | topi` | I tried running minicom on it |
| 18:46 | topi` | but yes, I'll try agetty as well, just to be sure |
| 18:47 | topi` | but I'm confused, from the schematics it seems UART3 goes to a connector called J35 |
| 18:48 | topi` | and UART2 goes to the mikrobus connector. so which one is mapped to ttymxc3? |
| 18:52 | jnettlet | uart1 is the serial console, uart2 is the bt chip, uart3 is to the mikrobus |
| 18:55 | topi` | right, UART3_TX_DATA goes to mikrobus connector |
| 18:56 | topi` | UART2_TX_DATA goes to that J35 |
| 18:56 | topi` | it's on the same schema sheet than the Mikrobus |
| 18:56 | jnettlet | That is Bluetooth on the microsom |
| 18:57 | topi` | "RS 485 or UART 3 pin hearder" |
| 18:58 | topi` | UART2 also seems to get routed to the mainboard |
| 18:58 | jnettlet | Yes that UART can be configured to do rs485 natively if the remote end has the proper hardware. It requires three upstream patches that allow half duplex mode to be enabled |
| 18:58 | topi` | you'd expect it to be consumed inside the uSOM since the wifi/bt module is there |
| 18:59 | jnettlet | Unless that module isn't populated |
| 18:59 | topi` | I already ordered a batch of the RS485 click boards so we start with those (requires no HW modifications) |
| 18:59 | jnettlet | Although I would need to look at the schematics to refresh my memory |
| 18:59 | topi` | I'm not 100% sure how we ordered these Gates.. maybe they did come with wifi/bt although the customer doesn't have a need for radios |
| 18:59 | jnettlet | We will get them sorted out |
| 19:01 | topi` | I guess the new Marvell boards consume most of your work hours these days? |
| 19:01 | topi` | probably the only thing in common is those A9 cores.. |
| 19:02 | jnettlet | more than that. And other projects not yet announced :) |
| 19:02 | jnettlet | But always time for the IMX6 users |
| 19:03 | jnettlet | It is still a very capable chip |
| 19:04 | topi` | I guess Solidrun is going to be on the IMX6 boat for a long time |
| 19:04 | topi` | this customer in particular is probably expecting to be able to get replacement HW for a decade or so |
| 19:04 | topi` | it's an industrial application |
| 19:06 | topi` | I always have doubts about these rs232<->usb cables. (actually this is a 3.3V uart -> usb cable) ... so I'll just try out the serial using another Hummingboard as the peer |
| 19:06 | topi` | as I know ttymxc0 works |
| 19:06 | topi` | on the older HB (it doesn't have another exposed UART; afaik) |
| 19:18 | topi` | odd! there's already a process latched onto /dev/ttymxc3 according to fuser |
| 19:18 | topi` | it's /usr/sbin/brcm_patchram_plus |
| 19:19 | topi` | brcm sounds like the bluetooth chip... but why is it on the /dev/ttymxc3 ? |
| 19:20 | vpeter | jnettlet: Ok, just so we (I) know it is not somewhere in spam. |
| 19:25 | topi` | only 6 interrupts on irq 61 even though I've been bombarding it with characters :/ |
| 19:26 | topi` | (the number 6 is also right after the boot) |
| 19:26 | topi` | I'd expect the ttymxc3 irq stays at 61 during the whole time: |
| 19:26 | topi` | [ 1.174955] 21f0000.serial: ttymxc3 at MMIO 0x21f0000 (irq = 61, base_baud = 5000000) is a IMX |
| 19:44 | topi` | jnettlet: I tried all rx/tx combinations with this cmdline: agetty /dev/ttymxc3 115200 |
| 19:44 | topi` | and no go. I did get a login out of ttymxc0, though |
| 19:45 | topi` | and this is despite disabling the brcm utility from systemd, so fuser shows a clean record for ttymxc3 |
| 19:46 | topi` | now, next steps? maybe I try to verify the mikrobus connector is properly soldered on this Gate |
| 19:47 | topi` | unfortunately I only have 1 Gate at home, the rest are at the office 400km away |
| 19:52 | topi` | rabeeh: is there any errata of the uarts on the HB Gate rev 1.2 board? I think my board is defective |
| 19:53 | topi` | it says "rev-1.2" on the silkscreen |
| 19:57 | jnettlet | topi` I alb stepping out footer a couples of hours now. I will test my board first thing in the morning and get back to you |
| 20:09 | topi` | no hurry, we still haven't got a date for meeting that customer, yet |
| 20:09 | topi` | but I'm trying to prepare as well as possible |