FS#49996 - [linux] vfat filesystem is not builtin, harder to maintain/debug uefi systems

Attached to Project: Arch Linux
Opened by Björn Linse (bfredl) - Friday, 08 July 2016, 16:51 GMT
Last edited by Doug Newgard (Scimmia) - Friday, 08 July 2016, 17:37 GMT
Task Type Feature Request
Category Kernel
Status Closed
Assigned To No-one
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 0
Private No

Details

Description:
Upgrading the linux kernel, when using an uefi boot manager requiring the kernel image to be on uefi (fat) partition, such as gummiboot, is harder than necessary. After the linux kernel gets upgraded, one would try to mount the uefi partition to copy over the kernel (either manually or by a automated script), which fails as the vfat module is not available. Now one could mitigate this somewhat by force loading "vfat" module on every boot. But this won't help in a situation where one forgot to copy the kernel or the automated copying failed for whatever reason, and the kernel version booted is inconsistent which the version in /lib/modules/. In this situation one would need to either temporarily downgrade the package from the cache (network is most likely not available in this situation), force ignoring any dependency conflicts, or manually extracting the "vfat" module and dependencies from the package.

I would suggest always building the vfat module into the kernel. The additional size (56K) seems to be a small price to pay to make fixing/debugging uefi systems a lot less painful.

Additional info:
* package version(s): linux-4.6.3-1


Steps to reproduce:
Run "pacman -Syu" which happens to upgrade linux
Immediately or after a reboot, try to mount the efi partition to copy over the kernel, which fails as the vfat module for the running kernel is not available.
This task depends upon

Closed by  Doug Newgard (Scimmia)
Friday, 08 July 2016, 17:37 GMT
Reason for closing:  Duplicate
Additional comments about closing:   FS#34518 
Comment by Doug Newgard (Scimmia) - Friday, 08 July 2016, 17:30 GMT
Sounds like a broken workflow to me.

Loading...