On Fri, Jun 10, 2011 at 07:20:58AM +0000, Emmanuel Dreyfus wrote: > On Thu, Jun 09, 2011 at 01:33:46PM -0400, Thor Lancelot Simon wrote: > > I don't love having an option for this. Is there a space or performance > > issue with making it (or extended attribute support in general!) the > > default? > > I potential problem: extended attribute autostart on the root filesystem > with a corrupted filesystem easily leads to a panic at mount time. This > cannot be recovered through single user mode. I suggest that if we make
eh? Don't corrupted filesystems cause a panic anyway? What is it about extended attributes that makes this more likely? Is this a jumping-though-a-NULL-pointer panic, or a "oops, something is inconsistent with ext. attrs. so I'll call panic()" panic? If it's the second, can the extended attributes be turned off at that point? eric
