help :-: i can't update to a new kernel in #asahi because the /boot is too full yet I only have one kernel installed

@meluzzy there is something wrong. you shouldn’t have kernel-uki-dtbloader to begin with.

What large (starting at 10MB) files do you have in boot?

Be careful not to remove any kernel-16k packages

@janne
this is /boot

@meluzzy there should be only variants ending in “+16K”

Can you deinstall “kernel-core” (you might need to specify the versions), that removes dependent kernel packages as well which is intended. Make sure no packages starting with “kernel-16k” will be removed.

@janne I dont really have any kernel-core only kernel-16k-core.
can't i just get rid of the fc43 ones?

UPDATE: deleted the fc43 ones and restarted, everything seems fine

@meluzzy @janne yeah you can delete old kernels just fine generally the boot partition doesn't have a lot of interdependent data
@meluzzy @janne does asahi work on x86 or is it only arm still
@hipsterelectron @janne I mean the idea is for it to remain aarch64 since its only purpose is to make the linux kernel run on new macbooks
@meluzzy @janne oh neat! thanks that makes sense
@meluzzy @janne was rly impressed at how much support was in the kernel for very complex peripherals like the touch bar when i went through every kconfig option a few months ago. i had had more difficulty than usual w linux on my 2020 x86 intel mac but it's sooooo nice how old hardware just becomes better over time in open source
@meluzzy @janne btw i found conflicting info regarding the arm memory model and the gcc cpu feature selection docs page is the best resource i have found for checking feature support for stuff like neon instructions—is there a canonical reference text everyone uses that describes how to check for cpu functionality on arm & lists all the variants of neon/simd? i'm working on experimental composable string search and particularly interested in instruction latencies but just a canonical list of features outside of the current gcc flags would be super useful
@meluzzy @janne SVE sounds cool but also sounds incompatible with fixed-size vectors and i wanna know if that's the future or if i can ignore it