-
jbkhrm.. does split /usr still work when booting? I'd like to construct an iso image with a small boot archive and then mount a larger /usr off the ISO image (vs. having one larger boot_archive), but don't know if that's something that'd still work
-
gitomat[illumos-gate] 18285 Want c23 uchar.h functions -- Robert Mustacchi <rm⊙fo>
-
tsoomejbk thats how OI is working (from install media) and smartos too. the catch is about getting /usr to be mounted - you have to have the needed bits in rootfs.
-
toastersonjbk: what tsoome said and you need the right scripts from slim_source in the exact right boot order disabling normal mounting or setup scripts.
-
richlowejbk: might have more luck with a linux-y root pivot
-
richloweas far as future proofing and such
-
jbkthe problem is because we still boot like it's 2005, making the boot archive too large causes problems
-
jbkalready have hit this with some AMD systems where their UEFI implementations don't leave enough contiguous memory free below 4G
-
jbkso we can't boot because our boot archives are too large
-
toastersonjbk: need this with another distro than Openindiana when there is no disk to keep / ? All other archives have been 200MB max
-
jbkit's not for openindiana
-
jbkbut the general AMD issue is true across every illumos distro i've tried
-
richloweis there enough memory _somewhere_?
-
jbkyeah... above 4GB
-
jbkplenty there
-
richloweand loader etc can't bounce and pull that off
-
richloweeugh
-
jbkno, tsoome has been working on it
-
tsoomeloader itself can, thats no problem;)
-
jbkbut it's basically unsupported by multiboot
-
jbkbut also apparently dboot seems to have some deep implicit assumptions about how physical memory is laid out
-
tsoomethe assumptions about memory layout are on the way - assuming I understand the problem correctly.
-
richlowebooting is hard, and we should stop
-
tsoomethat too:) well, the boot as such is not the problem. the problem is keeping some sanity while doing so;)