On 7/28/25 8:17 AM, GONG Ruiqi wrote:

On 7/26/2025 2:29 AM, Nayna Jain wrote:

quoted

On 7/17/25 8:29 AM, GONG Ruiqi wrote:

quoted

On 7/8/2025 4:35 AM, Nayna Jain wrote:

quoted

On 7/2/25 10:07 PM, GONG Ruiqi wrote:

quoted

...

Yes, IMA_ARCH_POLICY was not set. The testing was conducted on
openEuler[1], a Linux distro mainly for arm64 & x86, and the kernel was
compiled based on its own openeuler_defconfig[2], which set
IMA_ARCH_POLICY to N.

Thanks Ruiqi for the response.

It seems the main cause of the problem was that IMA_ARCH_POLICY config
wasn't enabled; and it sounds like you don't need IMA arch policies but
you do need the arch specific function to get the secure boot status.

In that case, removing IMA_SECURE_AND_OR_TRUSTED_BOOT config dependency
on IMA_ARCH_POLICY config and updating the corresponding help is all
that is needed.

I think it doesn't solve the real problems, which are: 1. the implicit
dependency of LOAD_UEFI_KEYS to IMA_SECURE_AND_OR_TRUSTED_BOOT, which
surprises people, and 2. what arch_ima_get_secureboot() does is
essentially a stand-alone function and it's not necessarily be a part of
IMA, but it's still controlled by IMA_SECURE_AND_OR_TRUSTED_BOOT.

I agree that adjusting Kconfig could be simpler, but breaking
IMA_SECURE_AND_OR_TRUSTED_BOOT's dependency to IMA_ARCH_POLICY doesn't
help on both. If that's gonna be the way we will take, what I would
propose is to let LOAD_UEFI_KEYS select IMA_SECURE_AND_OR_TRUSTED_BOOT,
which states the dependency explicitly so at least solves the problem 1.

Hi Ruiqi,

IMA_SECURE_AND_OR_TRUSTED_BOOT is already enabled by different 
architectures. Having LOAD_UEFI_KEYS select it would help only if 
IMA_ARCH_POLICY is also selected.

Thanks & Regards,

    - Nayna