I had it compiled 5 minutes ago, but I recently changed kernel's config and recompile it. This ebuild checks for BACKLIGHT_CLASS_DEVICE option. I have it enabled, but compilation fails with: /var/tmp/portage/app-laptop/nvidiabl-0.70/work/dkms_source_tree/nvidiabl-module.c:291:49: error: 'FB_BACKLIGHT_LEVELS' undeclared (first use in this function) So it seems it needs more than 1 kernel config to check in the ebuild, and also FB_BACKLIGHT_LEVELS is not actually a kernel config option, so I don't know which option defines this symbol to make it compile. Reproducible: Always
Update: I researched a bit on kernel's sources and found that FB_BACKLIGHT_LEVELS is actually defined when activating: FB_BACKLIGHT, so the fix to this will be to check in ebuild for FB_BACKLIGHT too.
More research: It seems that FB_BACKLIGHT is some kind of internal symbol which does not have a direct option under menuconfig. I found that it is enabled automatically if you compile any graphic adapter having a backlight option (for example nvidia framebuffer support (with backlight option turned on) I'll report it upstream requesting this option to be enabled with CONFIG_BACKLIGHT_CLASS_DEVICE (general one) instead of doing it with the others to have the possibility to install nvidiabl (and maybe others without problems) How about adding an ewarn to avoid people going crazy like myself?
I finally raised an upstream bug in order to see what they think about this subtle issue that drove me crazy. This is the URI: https://bugzilla.kernel.org/show_bug.cgi?id=40492
I'm actually pretty sure that the old check (BACKLIGHT_CLASS_DEVICE) was added specifically for this define, so they prolly changed it to FB_BACKLIGHT in some recent kernel. Gotta find out which version exactly and make the check do the right thing for both older and more recent versions (or depend on more recent if it's not too recent).
Uhm, actually, wait. The ebuild doesn't check for BACKLIGHT_CLASS_DEVICE, it checks for FB_BACKLIGHT: nvidiabl % grep CONFIG_CHECK nvidiabl-0.* nvidiabl-0.69.ebuild:CONFIG_CHECK="FB_BACKLIGHT" nvidiabl-0.70.ebuild:CONFIG_CHECK="FB_BACKLIGHT" This has been the case since: 18 Mar 2011; Christoph Mende <angelos@gentoo.org> nvidiabl-0.67.ebuild: Added check for FB_BACKLIGHT
Erm.. I rechecked it and yes, it checks for the correct option (sorry for my mistake!), ebuild seems correct and it seems upstream bug. Keep this open until a final response received from upstream, since the problem is not really here. But an ewarn is suggested anyways for having into account this issue.
Sorry, I don't see why this should be any bug at all. nvidiabl needs FB_BACKLIGHT, the ebuild checks for FB_BACKLIGHT and dies if it's not set. If at all, this is a portage bug because it didn't abort. Anyway, closing this as there's nothing that should be changed.
