[releng_6 tinderbox] failure on sparc64/sparc64

[M. Warner Losh]

In message: <86ek2iao05.fsf at xps.des.no>
            des at des.no (Dag-Erling Smørgrav) writes:
: "M. Warner Losh" <imp at bsdimp.com> writes:
: > If things have really changed, then we should change the default and
: > remove the kludges.  My main objection is the mismatch, not the actual
: > value.  Did you remove the kludges in the mk files at netapp to remove
: > the -fno-strict-alias?
: 
: Most of the kernel builds fine without -fno-strict-alias.  Some parts
: do not, and the committer responsible for introducing those errors
: added the kluges you mention to avoid having to fix his code.

There's many different files in the tree that fail to compile.  I find
this hard to believe.  A quick survey shows breakage in the following
areas:
	firewire, advansys, an, amr, ciss, hifn, isp, mly, pccbb, rue,
	ubser, ucom, msdosfs, nullfs, portalfs, udf, unionfs, ext2fs,
	reiserfs, cd9660, init_main, kern_exit, kern_fork,
	kern_malloc, kern_mtxpool, kern_synch, kern_thread, link_elf,
	sys_pipe, vfs_mount, bluetooth, gif, ufs, xfs, sbni,
	i386/machdep, mp_machdep, pmap, sys_machdep, imgact_coff, npx

I'm not sure about which committer you are talking about, but I can't
imagine that one person is responsible for all of that...

Warner