Afaic, latest 0.35-r1 and previous ebuilds don't use linux-mod
linux-mod wouldnt be an issue with "populate" as populate looks for portage owned files in /lib/modules/`uname -r`/. Can you attach the outoput of equery f shfs
This ebuild, if still maintained needs to be migrated to linux-mod. If you would like me to do it on your behalf, please let me know.
(In reply to comment #1) > linux-mod wouldnt be an issue with "populate" as populate looks for portage > owned files in /lib/modules/`uname -r`/. > > Can you attach the outoput of equery f shfs Yes, you're right, i've dug a little bit into module-rebuild and the problem is not related specifically to shfs (yet shfs should be ported to linux-mod so this bug is valid). The problem is i forgot to re-emerge shfs after last kernel update so the module was not in /lib/modules/<last kernel>. I wonder, is it right behavior for module-rebuild's populate ? i mean, even if i did not re-emerge the package lately, it's still letting in a broken package which was emerged at some point. what do you think about making it search through /lib/modules/.*\.k?o and deal with modules which are there but not under /lib/modules/`uname -r`: a) issuing a warning b) adding them if a certain param is used (also recompile modules not in use) c) adding them (since i guess param given at b) should be suggested by docs as default anyway) The problem with b) and c) would be if some package installs modules not needed and afaik only xorg does, so it could be masked either hard or by config. I can help you with the patch if time is lacking
> what do you think about making it search through /lib/modules/.*\.k?o > and deal with modules which are there but not under /lib/modules/`uname -r`: > a) issuing a warning > b) adding them if a certain param is used (also recompile modules not in use) > c) adding them (since i guess param given at b) should be suggested by docs as > default anyway) Correction: what do you think about making it search through /var/db/pkg/*/*/CONTENTS for /lib/modules/.*\.k?o and deal with modules which are there but not under /lib/modules/`uname -r`: a) issuing a warning "package may need to be re-emerged" b) adding package to be rebuilt if a certain param is used ("also recompile modules not in use") c) adding package to be rebuilt (since i guess param given at b) should be suggested by docs as default anyway)
not sure if this will answer all the questions re: populate you had, but populate is designed to populate the database from what is currently installed on the system. This is almost invalid for any user who has merged any modules recently (linux-mod) but has remained because of its ease of use.# In the future module-rebuild will give you the option to also execute scripts pre and post merge. I've not touched this in a little while though.
the sooner -r2 comes to stable the better. Looking at this the actual bug itself is a cantfix, since we wont be backporting this change to stable. Will make good note of the other part of this bug however.