On 2022-04-21, michal.lyszc...@bofc.pl <michal.lyszc...@bofc.pl> wrote: > --47wmzg5ty6ypgy6x > Content-Type: text/plain; charset=us-ascii > Content-Disposition: inline > Content-Transfer-Encoding: 7bit > > Hello Stuart, > Thanks for your reply, here is more data > On 2022-04-21 21:43:08, Stuart Henderson wrote: >> if you boot the upgrade kernel and drop to a shell, what do you >> get from this? >> >> cd /dev; sh MAKEDEV sd0 >> mount -t ffs -r /dev/sd0a /mnt >> ls /mnt > Sadly, nothing that would raise any flags or ideas in my brain, > everything seems to work fine > > upgrade# cd /dev; sh MAKEDEV sd0 > upgrade# mount -t ffs -r /dev/sd0a /mnt > upgrade# ls /mnt > .cshrc bsd dev sbin > .profile bsd.booted etc sys > altroot bsd.rd home tmp > auto_upgrade.conf bsd.sp mailwrapper.core usr > bin bsd.upgrade root var > upgrade# df -h > Filesystem Size Used Avail Capacity Mounted on > /dev/rd0a 3.5M 3.0M 451K 87% / > /dev/sd0a 3.9G 677M 3.0G 18% /mnt > >> it's worth seeing what "sysctl hw.disknames" says too > upgrade# sysctl hw.disknames > hw.disknames=sd0:dc999ef6267325df,rd0:a8c7c8e3bbaa0da7
That looks sane too.. >> (similar to what's used for the "is_rootdisk()" check in >> src/distrib/miniroot/install.sub) > Is there a way to run upgrade script with "set -x" globally? > I tried to run /bin/ksh -x /upgrade.sh, but it seems -x is > discarded in functions and I can only see debug up untile > +do_ugrade function call. > > Maybe replacing all "() {" with "() {\nset -x" will do the trick? > Not 100% sure but there are some functions that return output rather than just a return code and I think it may break those. Something more targetted in is_rootdisk might do though (set -x. maybe get rid of the redirect).