| 11:03 | topi`_ | jnettlet: question about the USB VBUS patches you supplied. Is it a lot of work to change the .dts so it works on the older Hummingboard (v1) ? |
| 11:04 | topi`_ | jnettlet: I need to liberate the Gate for another project now and I have several older HBs on my shelves :) |
| 11:22 | jnettlet | topi`_, it should just work on the HB1 as far as I know |
| 11:22 | jnettlet | if it doesn't let me know and I will look into it deeper |
| 11:23 | topi`_ | ok :) |
| 11:23 | topi`_ | I'll try |
| 11:23 | topi`_ | can I use the same kernel binary image on the HB1? (just copy the non-Edge .dtb to /boot)? |
| 11:24 | topi`_ | I think that's the whole idea behind devicetrees ;) but it's been too many years since I worked on the ARM kernel and back then we had to build the board support into the kernel |
| 11:24 | topi`_ | with those so-called "board files" |
| 11:26 | jnettlet | yes |
| 11:26 | jnettlet | really you should just have to move the sd card over |
| 11:26 | jnettlet | and it will work |
| 11:27 | jnettlet | you are using on of our debian images right? |
| 11:29 | topi`_ | yep |
| 11:29 | topi`_ | If I remember correctly, your patch touched just the hb_edge.dts so I need to replicate that change to the original hb dts |
| 11:29 | topi`_ | I'll check when I get back home :) |
| 11:30 | jnettlet | topi`_, let me look I think the original dts files were fine. |
| 11:30 | topi`_ | at least we have had hot water reliably for the past few weeks :) every morning the boiler has done its work to +80c during the cheap hours at night |
| 11:31 | topi`_ | the Gate is doing all the work controlling the electric supply to the boiler |
| 11:31 | jnettlet | topi`_, yeah the original device-tree files are fine. |
| 11:31 | jnettlet | great to hear! |
| 11:31 | topi`_ | so all that was missing, was adding that regulator file to /sys/class ? |
| 11:32 | topi`_ | userspace is important :) |
| 11:37 | jnettlet | the regulator file already existed but the state was read only. I believe this was because the maintainers wanted only the various device drivers to manage the regulators. |
| 11:38 | jnettlet | but usb had removed the ability to fully power off a usb node instead only leaving "auto" as the best power management option. |
| 11:38 | topi`_ | yeah, I noticed that |
| 11:39 | topi`_ | and the powering off of VBUS seems to be a neglected feature, most USB hubs shipping nowadays do not have it |
| 11:39 | jnettlet | all of this is a bit stupid. If someone wants to power off a regulator they should be able to. It is no different than hitting a switch on a device |
| 11:39 | topi`_ | yes |
| 11:39 | topi`_ | especially because USB is the easiest non-soldering way of controlling e.g. power sockets |
| 11:39 | topi`_ | just buy the cheap stuff from amazon and connect it :) |
| 11:40 | jnettlet | sure thing. |
| 11:40 | topi`_ | actually I could buy a cheap man's version of a "kill switch", e.g. a hardware watchdog, with this |
| 11:40 | topi`_ | perfect for the server closet |
| 11:41 | topi`_ | a Hummingboard, a usb-controllable poewr socket, and a suitable Watchdog application |
| 11:41 | topi`_ | that the server has to feed at some INET socketr |
| 11:41 | topi`_ | no bone in 10 minutes -> power cycle |
| 11:52 | jnettlet | topi`_, why not use the watchdog on the iMX6 |
| 11:52 | jnettlet | should be all working. Just enable support for it in systemd |
| 12:01 | KBme | jnettlet, and in u-boot? SPL? |
| 15:02 | topi`_ | jnettlet: what's the watchdog able to do? can it do a cold reset to a whole system? (how? manipulating the regulators?) |
| 15:02 | topi`_ | the problem with 4G modems (also with wifi cards) is that sometimes they're so busted that only a cold reset will help |
| 15:04 | topi`_ | if the +3.3V lines to miniPCIe can be switched on/off via a regulator, then I guess that's no longer a problem :) |
| 15:05 | jnettlet | KBme, there is support in u-boot for the watchdog but it is not enabled by default. |
| 15:06 | jnettlet | topi`_, the watchdog only does a reset. This means all the pads get reset to their defaults so most the regulators will get turned off. |
| 15:06 | topi`_ | aha, that's interesting |
| 15:06 | topi`_ | maybe I'll give it a shot and see what happens :) |
| 15:07 | jnettlet | the PCIe devices should get powered off because RST will get de-asserted |
| 15:07 | jnettlet | sorry PERST will get de-asserted |
| 15:11 | topi`_ | 4G modems probably aren't connected to any PCIe signals |
| 15:11 | topi`_ | the just use the USB wires I think |
| 15:11 | jnettlet | usually they are USB |
| 15:11 | topi`_ | and come with the assorted problems that ppl associate with USB :) |
| 15:11 | jnettlet | but the current boards have no regulators for that port of the hub....we are adding it in the next one |
| 15:11 | topi`_ | good news :) |
| 15:12 | topi`_ | I've had fairly good experiences with a specific Huawei 4G miniPCIe card - we've deployed it in tens of systems already and so far simple SW resets have been enough |
| 15:13 | topi`_ | but some older customers have Sierra modems which do lock up occasionally |
| 15:13 | jnettlet | we are working with Telit, but only 3G hardware. With their reset firmwares they have stabilized it quite well. |
| 15:13 | jnettlet | earlier versions used to wreak havoc on our usb hub. |
| 15:13 | topi`_ | scary how complex modern modems are :) |
| 16:12 | KBme | jnettlet, so when the watchdog resets the board I still have the boot bug where loading the kernel hangs forever. |
| 16:13 | KBme | I am not sure I've enabled the watchdog in u-boot, but it is enabled in the kernel. so now I open the watchdog device and let it reset the board, but the boot fails at loading kernel. |
| 16:13 | jnettlet | KBme, if you are at the point where the loading kernel hangs forever it means there is a kernel bug. |
| 16:13 | jnettlet | you need to enable earlyprintk and see where it is hanging |
| 16:14 | KBme | jnettlet, thanks. will do. |