dgilmore changed the topic of #fedora-riscv to: Fedora on RISC-V https://fedoraproject.org/wiki/Architectures/RISC-V || Logs: https://libera.irclog.whitequark.org/fedora-riscv || Alt Arch discussions are welcome in #fedora-alt-arches
iooi_ has joined #fedora-riscv
iooi has quit [Ping timeout: 264 seconds]
iooi_ is now known as iooi
bkeys has joined #fedora-riscv
<djdelorie> somlo: after the losetup command you should be able to run fdisk -l or parted on the loop0 device to list the partitions. If that works, you're at least that far
<djdelorie> you also have the option of using the file as-is as a physical disk in qemu, I suppose. It's an image of a physical hard drive
<somlo> djdelorie: didn't try to parted loop0, but there were loop0p[1..4] auto-detected, and it's *those* I could not mount
<somlo> also, there's three images of physical hard drives, so the question is which one goes where, when :)
<djdelorie> there are two sets of images: one big one for a microsd-only setup (write it to the microsd)
<djdelorie> and a small one/medium one for nvme: write the small one to the microsd, and the big one to wherever you want your root partition to be
<djdelorie> me, I wrote the one big one to both microsd and nvme, for consistency and redundancy
<djdelorie> (at least, on real hardware)
<djdelorie> the hifive needs the first two partitions of the microsd to contain firmware and a bootloader
<djdelorie> partitions 3 and 4 are /boot and /, and can be on microsd or nvme
<djdelorie> I don't know how to map all that to what qemu needs, but now you know what he images are, I hope ;-)
<djdelorie> as for mounting, you should at least be able to mount loop0p4 as it's ext4
<djdelorie> p3 may be ext4 or fat, fat may require code page support
<djdelorie> p1 and p2 are raw data and not mountable, I think
bkeys1 has joined #fedora-riscv
bkeys has quit [Read error: Connection reset by peer]
bkeys1 is now known as bkeys
bkeys has quit [Ping timeout: 245 seconds]
<davidlt[m]> p3 is ext too, there will be a FAT partition once we go EFI
<davidlt[m]> yes, p1 and p2 you cannot mount, these are RAW GPT partition containing binary data (firmware)
<davidlt[m]> bkeys: the firmware images goes to microSD card and the rootfs goes to NVMe
jcajka has joined #fedora-riscv
cyberpear has quit [Ping timeout: 268 seconds]
cyberpear has joined #fedora-riscv
<rwmjones> morning
jimwilson has quit [Ping timeout: 250 seconds]
jimwilson has joined #fedora-riscv
<tekkamanninja> morning
dgilmore has quit [Ping timeout: 264 seconds]
jcm has quit [Ping timeout: 264 seconds]
jcm has joined #fedora-riscv
dgilmore has joined #fedora-riscv
zsun has joined #fedora-riscv
bkeys has joined #fedora-riscv
zsun has quit [Remote host closed the connection]
jcm has quit [Ping timeout: 245 seconds]
jcm has joined #fedora-riscv
jcm has quit [Ping timeout: 245 seconds]
jcm has joined #fedora-riscv
<somlo> djdelorie, davidlt[m]: thanks! so, to recap, "...Unmatched.firmware.raw.img.xz" and "...Unmatched.rootfs.raw.img.xz" are complementary, and "Unmatched.raw.img.xz" contains both firmware and rootfs in a single image?
<djdelorie> yes. Although, the first set might be missing /boot; I haven't looked at that one myself
<somlo> ok, so initramfs unpacks to a cpio archive of 135MB, which then dictates the lower-bound RAM capacity of whatever one tries to boot this on :)
* somlo reaches directly for his genesys2 board, which has 1GB RAM...
<somlo> plan is to build a single blob (initramfs inside linux kernel as payload inside opensbi) that boots whatever's on the fourth partition (presumably the real root filesystem)
<somlo> it *almost* worked last time (back in the BBL era), but maybe it's simply a matter of throwing enough hardware at it to prevent systemd from hanging during boot...
<davidlt[m]> systemd doesn't like slow systems :)
jcajka has quit [Quit: Leaving]
<djdelorie> davidlt[m]: is there anything else I need to do to this hifive board before converting it to a builder and stashing it in the rack?
<davidlt[m]> djdelorie: probably that will need to wait for new disk image
<djdelorie> ok. Why?
<davidlt[m]> Better support ffor HW (hopefully) and some recent changes to Koji infra needs testing
<djdelorie> a "dnf update" won't bring those in?
<davidlt[m]> There is these autorpm stuff and that requires packages and configuration on worker
<djdelorie> (obviously, they won't bring in the firmware/uboot partitions)
<davidlt[m]> The packages are in, but so far I failed to get it configured properly
<davidlt[m]> Don't know + it requires manual configuration
<djdelorie> ok. For now I'm powering it off then, as the tiny fan is annoying
<davidlt[m]> otherwise packages fail like this: http://fedora.riscv.rocks/koji/buildinfo?buildID=196081
<davidlt[m]> python-rasterio-1.2.10-%autorelease
<davidlt[m]> the current disk image incl. all requires packages (AFAIK), but I couldn't get the configuration right thus I am getting some errors that require debugging
bkeys has quit [Ping timeout: 250 seconds]
bkeys has joined #fedora-riscv
<jimwilson> FYI I've got the experimental fedora image on my unmatched, working well so far, though I did manage to get one non-reproducible gcc ICE under load, same as I've seen on freedom-u-sdk
<jimwilson> I had to use a freedom-u-sdk image to copy the rootfs file to the nvme disk, the full sdcard image came up without any obvious nvme support