On 10/26/19 4:29 PM, DJ Lucas via lfs-dev wrote:
Any objection to adding pointers to BLFS on the kernel page about Linux firmware and cpio that might be used by make? The firmware has been a known issue for some time but was not included in LFS because it is covered in detail in BLFS (as well as limited to exactly what is needed so not have a bunch of unnecessary binary blobs laying around on the root filesystem). The cpio requirement (outside of an initrd) is relatively new for the built-in headers. See https://cateee.net/lkddb/web-lkddb/IKHEADERS.html for more info on that. We pulled in OpenSSL because of this, but that's not an esoteric requirement like above (and for most people, I think, the same for firmware blobs - I build in for ATI cards - but maybe not as we are seeing more of the NICs that need them now too). Personally, I just use a local copy of the book for jhalfs and add the entire firmware repo inline as a single install (http://www.linuxfromscratch.org/~dj/linux-firmware.svnstash <- if anybody needs for the same task) because I don't really care about the blobs or disk space, and I will do the same for cpio, but I think these should at least be mentioned in LFS, ideally IMO collapsing the two notes on that page with these.
A note mentioning firmware and a pointer is reasonable. For a first boot, networking is not an absolute requirement. There may be a better case for some video cards, but are there some that don't work at all without firmware?
The main goal at the end of LFS is to get get a login prompt and be able to log in to get a bash prompt. Anything else can be handled in BLFS.
-- Bruce -- http://lists.linuxfromscratch.org/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
