On Thu, Jun 16, 2011 at 05:10:18PM +0000, Emmanuel Dreyfus wrote: > On Thu, Jun 16, 2011 at 01:03:48PM -0400, Thor Lancelot Simon wrote: > > As you've observed, this poses an increased risk of filesystem corruption, > > and this code is not sufficiently robust to handle such corruption on > > system startup without crashing the kernel. > > With -o extattr it works fine, as the root is mounted without extended > attributes, and remounted with EA eanbled after fsck.
But since the extended attribute database isn't validated by fsck, the kernel can still suddenly crash at that point, no? > I understand that this implementation is weak, though. I do not have enough > time to address the problems, so I will not argue for adding UFS_EXTATTR > to generic kernels. I still think -o setattr can be be useful, as it > makes the thing a bit more usable. I know you didn't make the problems here -- you just inherited them. Thanks for working on this. Thor
