Re: [arch-ports] Booting i686 iso in VirtualBox
On Sat, Jun 10, 2017 at 12:25 PM, Mr Seth Price <ssterling@firemail.cc> wrote:
Is EFI enabled in the settings? I had a similar issue on my XPS 8700
till I enabled legacy BIOS. Beats me as to why, though. No, it was disabled. It will not even boot to the grub interface unless EFI is disabled. The screenshot was taken after selecting 'Boot Arch Linux (i686)' and the startup process failing. Additionally, the dual media seems to exhibit the same issue as the i686 only media.
Hi Tyzoid, thanks for testing. Can you confirm that the may iso still booted for you? Then it must be some "hidden" error on my end: The only thing that changed from may to june was the machine I build the iso on :-/ (and of course, the packages are newer - but still the official ones). I also uploaded a new iso just to double-check (if 2017.06.01 doesn't boot for you, 2017.06.11 should not boot, too). cheers, Erich On 10.06.2017 20:10, Tyzoid D wrote:
On Sat, Jun 10, 2017 at 12:25 PM, Mr Seth Price <ssterling@firemail.cc> wrote:
Is EFI enabled in the settings? I had a similar issue on my XPS 8700
till I enabled legacy BIOS. Beats me as to why, though.
No, it was disabled. It will not even boot to the grub interface unless EFI is disabled. The screenshot was taken after selecting 'Boot Arch Linux (i686)' and the startup process failing.
Additionally, the dual media seems to exhibit the same issue as the i686 only media. _______________________________________________ arch-ports mailing list arch-ports@archlinux.org https://lists.archlinux.org/listinfo/arch-ports
On Sun, Jun 11, 2017 at 3:29 PM, Erich Eckner <arch@eckner.net> wrote:
Hi Tyzoid,
thanks for testing. Can you confirm that the may iso still booted for you?
I haven't tested that one, I only grabbed the June one. I'm download that now to test it.
Then it must be some "hidden" error on my end: The only thing that changed from may to june was the machine I build the iso on :-/ (and of course, the packages are newer - but still the official ones).
I also uploaded a new iso just to double-check (if 2017.06.01 doesn't boot for you, 2017.06.11 should not boot, too).
Downloading. I'll send a status update with the progress testing this along with the 2017-05 one.
cheers, Erich
Thanks for your work getting this running! Cheers, Tyler
On Sun, Jun 11, 2017 at 3:29 PM, Erich Eckner <arch@eckner.net> wrote:
Hi Tyzoid,
thanks for testing. Can you confirm that the may iso still booted for you?
Yup, this worked. On to the next issue. I can't seem to get the pacman package to verify, even after instructing it to download your key. Can you verify that the package is not corrupt, as pacman claims? http://i.imgur.com/Th0Amsg.png
Then it must be some "hidden" error on my end: The only thing that changed from may to june was the machine I build the iso on :-/ (and of course, the packages are newer - but still the official ones).
I also uploaded a new iso just to double-check (if 2017.06.01 doesn't boot for you, 2017.06.11 should not boot, too).
Thanks, 2017.06.11 had booted, but had the same kernel panic as before.
cheers, Erich
Thanks, Tyler
Hi Tyler, On 11.06.2017 23:49, Tyzoid D wrote:
On Sun, Jun 11, 2017 at 3:29 PM, Erich Eckner <arch@eckner.net> wrote:
Hi Tyzoid,
thanks for testing. Can you confirm that the may iso still booted for you?
Yup, this worked.
Okay, now I have something to start at. I'll try to change the build process on the current machine as close to the one on the old one. This should be done within the next few days.
On to the next issue. I can't seem to get the pacman package to verify, even after instructing it to download your key. Can you verify that the package is not corrupt, as pacman claims? http://i.imgur.com/Th0Amsg.png
You're already using the new mirror, aren't you? In that case, you need to install archlinux32-keyring first on the installation medium (!). The problem was, that I had to include archlinux32-keyring as a pacman dependency, so it would get installed next to archlinux-keyring. But for this, I have to recompile pacman and sign it with my build key. On the other hand, the iso still uses the official packages - and especially the official pacman package without archlinux32-keyring.
Then it must be some "hidden" error on my end: The only thing that changed from may to june was the machine I build the iso on :-/ (and of course, the packages are newer - but still the official ones).
I also uploaded a new iso just to double-check (if 2017.06.01 doesn't boot for you, 2017.06.11 should not boot, too).
Thanks, 2017.06.11 had booted, but had the same kernel panic as before.
Thanks, Tyler
cheers, Erich
On Jun 12, 2017 12:48 AM, "Erich Eckner" <arch@eckner.net> wrote:
Hi Tyler,
Okay, now I have something to start at. I'll try to change the build process on the current machine as close to the one on the old one. This should be done within the next few days.
Sweet, thanks. I got your IRC message (though I was afk when you sent it), and I tested the archlinux-2017.06.12-i686.iso image, which also segfaults when starting up. Let me know if testing this on actual physical hardware would be helpful for your debugging process.
You're already using the new mirror, aren't you? In that case, you need to install archlinux32-keyring first on the installation medium (!).
Ah, I had thought archlinux-keyring would have been enough. It works after installing the archlinux32-keyring, but it seems to need to be installed manually via pacman -U. Pacman thinks that the package is corrupt unless the package is installed, bit of a catch-22. Here's a video documenting the my install process (using the archlinux-2017.05.01-i686.iso image): https://youtu.be/LEPKi9vGKK8
cheers, Erich
Thanks, Tyler
Hi,
Sweet, thanks. I got your IRC message (though I was afk when you sent it), and I tested the archlinux-2017.06.12-i686.iso image, which also segfaults when starting up. Let me know if testing this on actual physical hardware would be helpful for your debugging process.
Hmm, this is strange: Now it's exactly the same build script on the same machine as for the may iso :-/ I'm absolutely clueless what's wrong ... On the other hand, a segfault would be a little harsh for a wrongly composed set of packages ... maybe it's some conflict between the new kernel and the vm? :-/
You're already using the new mirror, aren't you? In that case, you need to install archlinux32-keyring first on the installation medium (!).
Ah, I had thought archlinux-keyring would have been enough. It works after installing the archlinux32-keyring, but it seems to need to be installed manually via pacman -U. Pacman thinks that the package is corrupt unless the package is installed, bit of a catch-22.
Its similar to archlinuxarm: There is the original archlinux-keyring package as well as a archlinux32-keyring (or archlinuxarm-keyring in the case of archlinuxarm). In the end, it will be a pacman dependency (in contrast to archlinuxarm!?), so nothing to bother about - but the current isos still use the official pacman package, so it needs to be installed explicitely.
Here's a video documenting the my install process (using the archlinux-2017.05.01-i686.iso image): https://youtu.be/LEPKi9vGKK8
This looks as I expect. cheers, Erich
So a quick update on where this is. We've figured out that the issue was caused by an update to the glibc package in the mainline repos. Pinning this back to a known good version allowed us to build working -i686.iso and -dual.iso images. These have been put in the master repo and should be syncing elsewhere shortly. Let us know if you have any issues with the new archlinux-2017.06.15-*.iso images. Thanks, Tyler
participants (3)
-
Erich Eckner
-
Tyler Dence
-
Tyzoid D