Skip to content
This repository was archived by the owner on Aug 5, 2022. It is now read-only.

Conversation

@pohly
Copy link
Contributor

@pohly pohly commented Feb 23, 2017

They were inconsistent and too small for future use cases.

Replaces PR #36 and needed for PR #19, which already had commit
13829d8.

The VFAT parition size specified in the DSK_IMAGE_LAYOUT no longer
matched the actual size used for images, and 15MB indeed is barely
large enough. Adding tools like cryptsetup to the initramfs causes it
to exceed that limit.

Instead of hard-coding a fixed size in the .wks.in file, now a
per-image variable is inserted instead. The size of the rootfs can
also be controlled, including not limiting it at all and thus adapting
the image size to the size of the rootfs. This will be needed for the
installer image which is larger than normal images.

The default is the same as before (30MB for the VFAT partitions, 4GB
total image size).

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
@mythi
Copy link
Contributor

mythi commented Feb 23, 2017

@pohly are the submodule updates necessary?

@pohly
Copy link
Contributor Author

pohly commented Feb 23, 2017 via email

30MB are not enough for an initramfs with cryptsetup included. Bumping
it up to the next power of two. The size of the rootfs gets adjusted
so that the full image fits into 4*10^9 bytes, i.e. "4GB" regardless
whether 1000 or 1024 is used as base. 64M are reserved for potential
partitioning overhead.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
@pohly
Copy link
Contributor Author

pohly commented Feb 23, 2017

Force-pushed updated commit.

@mythi
Copy link
Contributor

mythi commented Feb 23, 2017

test this please

@ipuustin
Copy link
Contributor

@pohly I also keep committing the submodules unintentionally (git commit . is the main cause). I was thinking of a commit hook to warn about it, but I don't know if that would interfere too much with the legitimate submodule updates.

@pohly
Copy link
Contributor Author

pohly commented Feb 23, 2017

@ipuustin if the change for the submodule was a bit more obvious in "git diff" output perhaps it would be easier to spot the mistake - at least I tend to check my change that way before committing. But I've not found a way to get more than the two-line change with old and new hash, which is just too easy to miss.

@mythi
Copy link
Contributor

mythi commented Feb 24, 2017

+1

@jlaako Are you OK?

@jlaako
Copy link
Contributor

jlaako commented Feb 24, 2017

@mythi yes, +1

@mythi mythi merged commit 0537c05 into intel:master Feb 24, 2017
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants