> On Mar 26, 2015, at 9:04 PM, Rui Paulo <rpa...@me.com> wrote: > > On Mar 26, 2015, at 19:35, Warner Losh <i...@freebsd.org> wrote: >> >> Author: imp >> Date: Fri Mar 27 02:35:25 2015 >> New Revision: 280728 >> URL: https://svnweb.freebsd.org/changeset/base/280728 >> >> Log: >> Categorize certain kernel builds as being broken in certain places. >> >> Differential Revision: https://reviews.freebsd.org/D2011 >> >> Modified: >> head/sys/conf/kern.opts.mk >> >> Modified: head/sys/conf/kern.opts.mk >> ============================================================================== >> --- head/sys/conf/kern.opts.mk Fri Mar 27 02:35:11 2015 >> (r280727) >> +++ head/sys/conf/kern.opts.mk Fri Mar 27 02:35:25 2015 >> (r280728) >> @@ -75,6 +75,34 @@ BROKEN_OPTIONS+= EISA >> BROKEN_OPTIONS+= OFED >> .endif >> >> +# Options that cannot be turned on this architecture, usually because >> +# of compilation or other issues so severe it cannot be used even >> +# on an experimental basis >> +__ALWAYS_NO_OPTIONS= >> + >> +# Things that don't work based on the CPU >> +.if ${MACHINE_CPUARCH} == "arm" >> +__ALWAYS_NO_OPTIONS+= CDDL ZFS >> +.endif > > This isn't entirely true. With a properly tuned ARC and kmem size, ZFS seems > to work on a BeagleBone. There have been reports on the freebsd-arm mailing > list this month. Based on my understanding of your previous commit, it looks > like this can't be overridden which is a problem.
Yea, the ships passed in the night. I’ve had these changes in my tree for a while. However, It kinda can be overridden by compiling the modules directly. Since the actual module’s Makefiles don’t depend on this, I think we’re OK. Can you test building a kernel on the BBB you have and let me know if it produces working ZFS? Warner
signature.asc
Description: Message signed with OpenPGP using GPGMail
