| ham5urg | Never used an ARM-cpu. My main-issue is that I don't know about its way of booting. Maybe it does not have something like an BIOS/EFI. So the boot-manager (if it can be called like that) would sit one more layer downward. | 23:07 |
|---|---|---|
| ham5urg | Can you light the boot process a bit up, before I would order a board? | 23:07 |
| ham5urg | If there is an ATX based ARM board, I would take a look at it. | 23:08 |
| c0rnelius | I can't speak for ever arm board ever, but basically you need u-boot. Each SoC uses a different partition offset and they are pretty much all flashed differently to the img. | 23:10 |
| c0rnelius | ever/every* | 23:10 |
| c0rnelius | Some use arm trusted firmware, amlogic uses boot fips provided by the vendor. Which can be created ur self, but are still in the end proprietary. | 23:11 |
| gnarface | ham5urg: yea they all do it a little differently but basically what happens is it's hardwired to look at a specific byte-offset or range of bytes for something executable, then run it. that thing is almost always these days either u-boot or a patched fork of it, which then initializes the hardware the way you'd expect a PC BIOS to, then hands off to Linux - once Linux is up it's pretty much the same from there out | 23:13 |
| gnarface | ham5urg: there's been some marginally successful efforts to shoehorn some EFI and Grub crap in there between u-boot and the linux kernel so that RedHat's disk images can be universal but it's completely unnecessary and possibly insane to go through that trouble | 23:15 |
| gnarface | ham5urg: s/executable/bootable/ | 23:15 |
| c0rnelius | I would say the easiest u-boot to work with is Allwinner as you don't even need to offset the partition to flash it and its super easy to compile. | 23:15 |
| gnarface | yea the pine64 allwinner stuff should work with the mainline u-boot now, no extra patches | 23:16 |
| gnarface | broadcom's rasberry pi crap needs something super proprietary | 23:16 |
| gnarface | most the rest of the stuff just needs one or two patches | 23:16 |
| ham5urg | I understand, so u-boot is something similar to an BIOS/EFI? Maybe without GUI but that would be even better. | 23:16 |
| c0rnelius | Not a fan of EFI myself so I don't bother with it. Its like creating an extra step just to make one for no real reason. | 23:16 |
| gnarface | ham5urg: it's a lot more like if someone tried to replace the BIOS/EFI entirely with a fork of the linux kernel | 23:17 |
| ham5urg | great | 23:17 |
| c0rnelius | You can create a simple gui type menu without EFI, using extlinux.conf | 23:17 |
| gnarface | ham5urg: it has the same base functionality but it can actually do a lot more if you look at the full breadth of the build options it's basically a whole OS unto itself | 23:17 |
| c0rnelius | BUt even that is kind of over kill. No one is usually installing and boot multi kernels on an arm board. | 23:18 |
| gnarface | usually | 23:18 |
| gnarface | though it's becoming popular with pinephones | 23:18 |
| gnarface | there's a multi-boot fork of u-boot for pinephones called p-boot | 23:18 |
| gnarface | (specifically it adds a graphical menu that you can control with volume buttons) | 23:19 |
| c0rnelius | There is also spi | 23:19 |
| c0rnelius | which hardkernel uses | 23:19 |
| ham5urg | I'm curious why there are no amd64-boards with uboot and removed efi? Is this a non-technical decision? | 23:19 |
| gnarface | i don't think there would be any demand | 23:19 |
| gnarface | what is more likely to happen is eventually we'll start seeing more arm64 boards with a bios | 23:20 |
| ham5urg | I would be the first customer to replace all efi | 23:20 |
| gnarface | the omission of the bios isn't a nextgen/upgrade type thing; is primarily a cost-saving effort by the manufacturers because clock chips cost | 23:20 |
| ham5urg | I don't understand what a clock chip has to do with it? I know that simple USB-clock-chips are a bit expensive. | 23:22 |
| gnarface | i think that's the core feature they're cutting out | 23:22 |
| ham5urg | And that a amd64 is running on false time after a couple of days | 23:22 |
| ham5urg | ARM cpu's have a built in clock chip? | 23:22 |
| gnarface | i don't think they all have one | 23:23 |
| ham5urg | ok | 23:23 |
| gnarface | amputated along with the bios and the non-volatile ram | 23:23 |
| gnarface | big complication actually; if you don't run ntpd on them you lose massive time while unpowered | 23:24 |
| gnarface | not a problem for phones and other mobile devices this stuff was originally invented for; nobody expected consumers to use any of this stuff without a battery | 23:24 |
| ham5urg | Yes, I have a rs232 clock to circumvent that issue. clock mouse I guess it is called. | 23:25 |
| ham5urg | I was looking for a tiny usb clock, simmilar to the wireless-keyboard-dongle but never found one (for laptop use) | 23:26 |
| gnarface | pine64 makes a smarwatch | 23:27 |
| gnarface | smartwatch* | 23:27 |
| gnarface | you could probably make it do that | 23:28 |
| gnarface | hell, you could probably just host a network time server on it | 23:28 |
| gnarface | haha | 23:28 |
| gnarface | i'm having a lot of fun with pine64 stuff lately | 23:28 |
| ham5urg | Maybe even on their pencil :D | 23:28 |
| ham5urg | Yeah, they are building a successful brand. | 23:29 |
| gnarface | i really want to just get a massive stack of them and build a giant cluster but i have absolutely no purpose for it | 23:29 |
| gnarface | of the bare soc boards they sell | 23:30 |
| ham5urg | Calculate the ROI for XMR :D | 23:30 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!