On May 20, 2014, at 3:20 PM, John-Mark Gurney <j...@funkthat.com> wrote:
> Warner Losh wrote this message on Tue, May 20, 2014 at 15:16 -0600: >> >> On May 20, 2014, at 3:14 PM, John-Mark Gurney <j...@funkthat.com> wrote: >> >>> Andrew Turner wrote this message on Sun, May 18, 2014 at 20:56 +0100: >>>> On Sat, 17 May 2014 20:31:34 +0000 (UTC) >>>> Warner Losh <i...@freebsd.org> wrote: >>>> >>>>> Author: imp >>>>> Date: Sat May 17 20:31:34 2014 >>>>> New Revision: 266349 >>>>> URL: http://svnweb.freebsd.org/changeset/base/266349 >>>>> >>>>> Log: >>>>> The time is not yet ripe to break the lack of dependencies between >>>>> src/sys and the rest of the tree for builds. >>>>> o eliminate including bsd.mkopts.mk for the moment in kern.opts.mk >>>>> o No need to include src.opts.mk at all anymore. The reasons for it >>>>> are now coverted in sys.mk and src.sys.mk. >>>> >>>> This breaks ARM kernel builds as MK_ARM_EABI is undefined, at least on >>>> 9.x. The below patch fixes it for me. >>> >>> But isn't armeb's ABI OABI on 9? so defaulting this to yes would change >>> the ABI as you build on 9.x wouldn't it? Or am I just confused by your >>> throwing in 9.x into the mix? >> >> First, this is current only. >> >> Second, I?m not changing the ABI on 9. >> >> Third, building 9.x armeb binaries is unaffected by the host compiler and >> build environment. It will continue to be what it was last week or last >> month independent of my changes. > > Oh, I think Andy was complaining about building -HEAD on 9.x, not about > building 9.x ARM... Then this change makes more sense and as you said, > isn't changing the ABI on 9... > > Ok, if that's the case, the confusion has been cleared... Yea, that issue has been corrected. The new build system didn’t define something, and 9.x was incidental… Warner
signature.asc
Description: Message signed with OpenPGP using GPGMail
