For some reason, after having emerged said package I see the following directories which are all empty: /usr/src/linux-3.9.2-gentoo/b /usr/src/linux-3.9.2-gentoo/b/Documentation /usr/src/linux-3.9.2-gentoo/b/Documentation/fb /usr/src/linux-3.9.2-gentoo/b/include /usr/src/linux-3.9.2-gentoo/b/include/linux /usr/src/linux-3.9.2-gentoo/b/drivers /usr/src/linux-3.9.2-gentoo/b/drivers/gpu /usr/src/linux-3.9.2-gentoo/b/drivers/gpu/drm /usr/src/linux-3.9.2-gentoo/b/drivers/gpu/drm/radeon /usr/src/linux-3.9.2-gentoo/b/drivers/video /usr/src/linux-3.9.2-gentoo/b/drivers/video/console /usr/src/linux-3.9.2-gentoo/b/drivers/char /usr/src/linux-3.9.2-gentoo/b/drivers/char/kcopy equery tells me that these belong to gentoo-sources; example: # equery belongs /usr/src/linux-3.9.2-gentoo/b/Documentation/fb * Searching for /usr/src/linux-3.9.2-gentoo/b/Documentation/fb ... sys-kernel/gentoo-sources-3.9.2 (/usr/src/linux-3.9.2-gentoo/b/Documentation/fb) I realise that this probably isn't a big deal, they are empty directories after all. But a fluke is a fluke no matter how trivial so I had to report it. Perhaps a bug snuck into the merging of patches..?
Multiple kernel versions have been doing this, this must either be a shared patch between the multiple branches or an error in the eclass; we'll try to look into this.
This isn't a problem as long as these directories contain no files; for these directories to contain no file, I'm using a QA check that checks for files in that folder to ensure patches aren't wrongly applied to an incorrect depth. If I see them again, I'll try to see if I can figure out which patch does this; but in general, there is no need to worry.
No problem. Must be a patch because the weird thing is, it is directory /b/.