James Bottomley wrote: > > The fatal flaw is that the C/H/S information is used by the msdos disk > label. Change it on the fly and existing partition tables don't say > what they previously said (this isn't fatal, since the CHS geometry is > actually encoded in the partition table, so as long as you don't > repartition it's preserved). The other flaw is that the 255/63/X C/H/S > is carefully crafted to let us limp to a 0.5TiB volume size with dos > labels. If we change that to 255/56/X we can only limp to about 480GiB > meaning you won't be able to put a dos label on the new 0.5TiB SATA > drives ... or rather you will (since we just silently truncate), but > you'll be surprised to learn your disk shrunk ... > There is nothing about C/H/S which requires us to use stupid boundaries. All it is is a convention, which happens to be dating back from early versions of the MS-DOS boot sector requiring that IO.SYS was contiguous in a single track. > The problems being described all come down to 4k sector disks pretending > to be 512b sector ones so as to remain "compatible" in windows. In this > case it's hard to prevent a user doing the wrong thing since they > naturally use msdos labels and end up with the misalignment. Note, we > can't even necessarily tell users to turn off windows compat mode on the > disks because most BIOSs aren't 4k ready and so can't boot properly > unless it's enabled. > > It's all a bit of a mess. RAID, flash, etc. have already been doing this for a long time. This isn't anything new. -hpa -- H. Peter Anvin, Intel Open Source Technology Center I work for Intel. I don't speak on their behalf. -- To unsubscribe from this list: send the line "unsubscribe util-linux-ng" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
