On Wed, Jan 13, 2016 at 8:33 AM, Eric Griffith <egriffith92@xxxxxxxxx> wrote: > > On Jan 12, 2016 15:03, "Josh Boyer" <jwboyer@xxxxxxxxxxxxxxxxx> wrote: >> >> On Tue, Jan 12, 2016 at 2:08 PM, Mattia Verga <mattia.verga@xxxxxxxxxx> >> wrote: >> > Il 07/01/2016 20:30, Tomasz Torcz ha scritto: >> >> >> >> On Thu, Jan 07, 2016 at 01:33:22PM -0500, Josh Boyer wrote: >> >>> >> >>> Hello, >> >>> >> >>> The 4.3.3 kernel has been pushed to updates-testing for F23. As of >> >>> right now, it has a +12 karma score. Given that it is a major release >> >>> rebase, we're going to wait at least a few days to see how it fares. >> >>> >> >>> If you are so inclined, testing would be appreciated. As usual, >> >>> please give karma as appropriate but we would appreciate it if you >> >>> only give negative karma for new, not reported issues and with a bug >> >>> link associated. If a bug is fixed, we have marked it as such. If it >> >>> isn't, we haven't and giving negative karma for those known issues >> >>> simply prevents fixes from getting into the hands of other users. >> >>> >> >> Please note there seem to be a btrfs regression in since 4.3: >> >> namely fstrim could discard beginning of the disk, removing the >> >> bootloader. This commit fixes the issue: >> >> >> >> >> >> >> >> http://git.kernel.org/cgit/linux/kernel/git/fdmanana/linux.git/commit/?h=integration-4.5&id=7b6cb6618b45bb383f9336ec89df5f1f31f9935b >> >> >> > So, is it safe to install kernel 4.3.3-300? I see it's now in stable >> > repo, >> > but I think it doesn't carry the commit above... >> >> The commit above isn't in any released kernel, Fedora or upstream or >> otherwise. It was only added to the main btrfs maintainer's tree >> about a day ago. >> >> As far as I understand the bug, it only happens if you have the >> bootloader installed in the root partition (not a separate /boot) and >> you run an fstrim on the whole partition (fstrim /) either manually or >> from a system script or such. I suppose it would possibly happen if >> you have the bootloader installed in a btrfs /boot partition and ran >> fstrim on that too. However, Fedora doesn't support booting from a >> btrfs filesystem at the moment and typical installs create a separate >> /boot partition that is ext4. I may have some details wrong and would >> gladly be corrected if so. > > Can confirm. Just head to reinstall an F23 system. Automatically generated > partitions (Encrypted btrfs raid0) gave me an ext4 /boot It takes a lot of gymnastics to get Fedora into this situation, and it can't be done via the GUI installer and maybe not even with kickstart. Whereas on openSUSE, where the bug was found, it's their default layout at install time. First they don't step on the bootloader in LBA 0 by default, and new installs use parted's jump code. So no matter what, the MBR bootloader code jumps to the VBR of the partition with an active bit set. The Btrfs VBR is huge, 64KiB, which is why it's supported as a GRUB bootloader pad, and that's where openSUSE embeds GRUB whether on MBR or GPT disks. So anyone with a 92 ways to boot Linux wall poster needs to update it to 93. -- Chris Murphy -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx http://lists.fedoraproject.org/admin/lists/devel@xxxxxxxxxxxxxxxxxxxxxxx
