siteinfo/insane: More n32 fixups
authorPeter Seebach <peter.seebach@windriver.com>
Thu, 21 Feb 2013 17:40:22 +0000 (11:40 -0600)
committerRichard Purdie <richard.purdie@linuxfoundation.org>
Mon, 25 Feb 2013 13:58:19 +0000 (05:58 -0800)
commitb7b2a671fd727b3879f30c29ba8eb5255ab2e168
treec44696f9f9024793152c5d0557405a6e05a316c2
parent93e5597fc386d49b556b8a13f1a42f5c6b5dc9f5
siteinfo/insane: More n32 fixups

For perfectly sound reasons, n32 MIPS is a mips64 subset denoted by an
ABI change, thus, "mips64-vendor-linux-gnun32" rather than
"mips-vendor-linux-gnu". A previous change had fixed up insane.bbclass
to recognize these, which mostly worked, but left SITEINFO_BITS set to
64.

Since bit-32 is processed first, and there are specific checks for
linux-gnun32, modify the mips64-linux-gnun32 lines to specify "bit-32",
so things that check SITEINFO_BITS get the right answer. Also, drop
the mips{,el}-linux-gnun32 lines, because that's not a valid combination;
n32 only makes sense for mips64.

Also, the insane.bbclass change spelled "mips64el" as "mipsel64", which
no one noticed because no one's using little-endian n32, apparently.

(From OE-Core rev: 0c243a550f5a0e539823fe168bcc7b2bbb9f19e1)

Signed-off-by: Peter Seebach <peter.seebach@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
meta/classes/insane.bbclass
meta/classes/siteinfo.bbclass