If you boot the boot-nonfree isolinux image and specify an HTTP install, stage1 immediately produces an error saying that it can't load drivers. The driver mentioned is atl1c, which would be the correct driver for the wired NIC on the laptop. There are also previous messages on tty1 indicating that ehc1 drivers could not be found. It never gets to the point of prompting for a selection of the wired or wireless NIC, which has worked very well in the past. Should be very easy to reproduce. Reproducible: Steps to Reproduce:
It worked when I tried a fresh install in relation to https://bugs.mageia.org/show_bug.cgi?id=17217 on Dec. 21, but fails today.
The first time the failure was seen was January 1.
Can you provide more info (eg: screenshots of tty2-4)?
Summary: Stage 1 network install is totally borked => Stage 1 network install failed to load atl1cSource RPM: (none) => drakx-installer-imagesKeywords: (none) => NEEDINFOCC: (none) => thierry.vignaud, tmb
This appears to be fixed in today's stage1. However, today's stage2 seems to have broken support for touchpads as mice. I'll enter another bug for that.
Status: NEW => RESOLVEDResolution: (none) => FIXED
This is broken again in today's stage1. The grub menu.lst entry to reproduce this is: title isolinux kernel (hd0,6)/cauldron/x86_64/isolinux/x86_64/vmlinuz BOOT_IMAGE=isolinux acpi=on resume=/dev/sda5 vga=791 initrd (hd0,6)/cauldron/x86_64/isolinux/x86_64/all-nonfree.rdz with obvious changes for your own environment. In mine, (hd0,6)/cauldron/x86_64 /isolinux/ is rsync'd from the mirror directory of the same name (there is also a cauldron/i586/isolinux/). The error is immediate. The curses prompt for install type appears, and the first few lines of tty1 are overwritten with messages about ehci drives being missing. If you respond to the install type with any network-based option, then you get the error about atl1c. I don't think that this has anything to do with atl1c. I think all-nonfree or possibly isolinux/vmlinuz is getting built now and then with several missing drivers. Or possibly not getting built at all: [ftg@ftgme2 x86_64]$ ls -l total 75180 -rw-r--r-- 1 ftg ftg 45311100 Jan 21 17:44 all-nonfree.rdz -rw-r--r-- 1 ftg ftg 27224392 Jan 29 05:24 all.rdz -rw-r--r-- 1 ftg ftg 4441824 Jan 29 05:24 vmlinuz Why would all-nonfree not have the same date as all ?
Resolution: FIXED => (none)Status: RESOLVED => REOPENED
Summary: Stage 1 network install failed to load atl1c => Stage 1 all-nonfree is sometimes not built
(In reply to Frank Griffin from comment #5) > This is broken again in today's stage1. > Why would all-nonfree not have the same date as all ? Because it need to be submitted separately, and since adding arm support the delay for rebuilding the nonfree can take a while, and sometimes I simply forget to submit it... So thie "bug" is simply that...
OK, I'll bite. Why is all-nonfree separate from all if it needs to be rebuilt whenever all is ?
Because it's nonfree? :)
because "all" is its submitted to core, and "all-nonfree" is submitted to nonfree to pick up nonfree firmwares, something that is not available in the core build... Anyway I think blino and pterjan have some fixes to get it better...
Ah, OK. I learn something new every day. It never occurred to me that the entire mirror tree wasn't available to packages being built. Do you want me to close this or leave it open for blino and pterjan ?
(In reply to Frank Griffin from comment #10) > Ah, OK. I learn something new every day. It never occurred to me that the > entire mirror tree wasn't available to packages being built. yeah, it's intentional as we want "Core" to be selfcontained, so no nonfree/tainted stuff available during Core builds... When we build nonfree, we add core + nonfree medias... when we build tainted, we add core + tainted medias... > Do you want me to close this or leave it open for blino and pterjan ? Well, I will try to ensure keeping the builds in sync now, probably by simply bumping rel for nonfree until buildsystem is fixed... A 4.4.1 kernel is currently building, so I'll get them in sync after that
Assigning to packagers collectively since it was still assigned to Bug Squad and I don't know who to assign it to. Adding blino and pterjan in CC.
Keywords: NEEDINFO => (none)CC: (none) => mageia, pterjan, stormiAssignee: bugsquad => pkg-bugs
Not an installer bug per se, more a BS issue
CC: (none) => sysadmin-bugsComponent: Installer => Release (media or process)Assignee: pkg-bugs => tmb
Source RPM: drakx-installer-images => (none)
Is there any reason to leave this open ? I haven't seen the problem for quite a while, so either the BS was fixed or Thomas is submitting nonfree reliably.
Buildsystem is fixed so same versioned builds can be submitted to core/nonfree/tainted at the same time
Status: REOPENED => RESOLVEDResolution: (none) => FIXED