On Jan 2, 2009, at 10:14 PM, M. Warner Losh wrote: > In message: <20090103033543.GB77475_at_dragon.NUXI.org> > "David O'Brien" <obrien_at_freebsd.org> writes: > : On Fri, Jan 02, 2009 at 11:47:57AM -0700, M. Warner Losh wrote: > : > In message: <20090102091334.GA41230_at_dragon.NUXI.org> > : > "David O'Brien" <obrien_at_freebsd.org> writes: > : > : Before 'fsck' would read the lable for the FS type. That has > changed and > : > : thus you cannot just 'fsck /dev/ad1s1d' anymore. So the FS > type in fstab > : > : must be accurate. > : > > : > Why did that change? I routinely have disks that aren't in my > : > /etc/fstab that I mount and this is a pain in the backside. > : > : Due to r186240 which: > : Make gpart the default partitioning class on all platforms. > : > : Seems GEOM_PART_BSD does not like labels that GEOM_BSD did. > > Then why the change? Shouldn't we make it like them for > compatibility? David's statement is incorrect. They like the same labels for all practical purposes. The problem at hand here is that fsck(8) & newfs(8) ask of GEOM_BSD what the partition type is. This means 2 things: 1. Any platform that doesn't use the BSD disklabel by default needs to have a proper /etc/fstab and is not behaving the same as i386. 2. Switching to GPT as the default partitioning scheme yields the same problem. As I said in an email to arch_at_, it's good to query the partition type to determine what to do when more specific information is missing (running fsck(8) vs fsck_ffs(8)). But it wasn't something that was there at the switchover point. FYI, -- Marcel Moolenaar xcllnt_at_mac.comReceived on Sat Jan 03 2009 - 16:32:46 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:39 UTC