Bug 2091986

Reproduction steps:
1. Run installer on a laptop with flaky efi that won't accept new boot entries.
2. When installer asks if I want to continue installation despite that, say yes.
3. Finish installation, reboot machine.
4. Have grub2 run correctly, but with no menuentries to boot into (except "UEFI Firmware Settings"), since no configs under `/boot/loader/entries` are present.

Expected result:
1. Even when installer encounters errors when editing efi boot entries, the kernel boot config under `/boot/loader/entries` should've been generated to increase the chance system will be bootable.

Context:
I have an old Acer Spin 1 SP111-32N on which I've just recently installed fedora 36 (kde version). Unfortunately the laptop has a known flaky efi that will not let efibootmgr set new entries correctly[1].

[1] For comparison here's the gymnastics needed to get ubuntu installed on the machine – https://gist.github.com/volcanoauthors/c7c9480dc78f1ba55d589d2125db4832

It appears to me that the installer in effect assumes that if it can't edit efi entries, then that means nothing will work. That's too pessimistic. On my particular machine even with the boot entry not set correctly (it says `Unknown Device`), the Acer Insyde H2O BIOS has a reasonable fallback in that it still looks at the emmc drive, finds the efi partition and runs whatever's present in `EFI/BOOT`. And that fallback is fully sufficient to get a working system as long as there are bls configs for grub to read.



Workaround:

If you found this via google and are hoping for a solution:
1. Once the installation completes, don't shut down the installer.
2. Open a terminal window, and `sudo chroot /mnt/sysroot` (I think it was /mnt/sysroot…; you can run `df -h` and look at last few entries to verify where your disk is mounted)
3. Regenerate kernel configs with `kernel-install add`. For me with the kernel I have in `/boot` the command looks like: `kernel-install add 5.16.20-200.fc35.x86_64 /lib/modules/5.16.20-200.fc35.x86_64/vmlinuz`. Or you can just `dnf reinstall kernel-core`, but that takes longer.
4. `cat /etc/fstab` and take note of the UUID of the `/` mount point. Say mine is `e8b683a6-f15d-4946-912e-3ca373dd6820`
5. Go to `/boot/loader/entries` and edit the config that's not the rescue to make sure the line starting with `options` is correct. Mine wasn't.
5a. For me the correct line would be (without disk encryption): `options root=UUID=e8b683a6-f15d-4946-912e-3ca373dd6820 ro rootflags=subvol=root`
5b. Or with disk encryption: `options root=UUID=e8b683a6-f15d-4946-912e-3ca373dd6820 ro rootflags=subvol=root rd.luks.uuid=luks-745d007e-e2f5-4b3f-949a-0459faa09099`
5c. You can find the `rd.luks.uuid=...` bit in `/etc/default/grub`.
6. Close the terminal window and now you can complete the installation.