Messages in this thread | | | Date | Wed, 10 Jun 1998 15:20:25 +0200 (MET DST) | From | Cyrille Chepelov <> | Subject | Re: MO better!! |
| |
On Wed, 10 Jun 1998, Harvey Fishman wrote:
> Where I apparently ran into problems was trying to read a disk with a > partitioned NTFS file system made under NT with Linux. It apparently > cannot do that. And then I compounded the problem by looking at the disk > with Linux fdisk, which showed me the apparent size of about 150 MB. ^^^^ About 150Mb * 4 = about 600 Mb (the true capacity of these disks -- 640 is with "commercial" megabytes (10^6 instead of the One True Definition Of Mega (1024^2)). 512 (the 'usual' sector size for most media) * 4 = 2048 (the MOD-640 sector size). Is that a coincidence, or a bug in NT (which one day will become "law") ? Interestingly, Win9* has no problem reading NT-partitioned disks (as long as they're formatted in VFAT), so it must be following the same definition. Looks like there's a mismatch between Win* and Linux (and perhaps the tiny rest of the world) regarding the unit used in parition tables to describe the disk. My first bet was that fdisk counted in 512-byte blocks where Win* were counting in actual sector sizes, but after looking into the relevant code in Linux' kernel, I got really puzzled, it seems to care about actual sector size and count in that units... only seems, alas ! (although a little pointless, partitioning a cartridge into several chunks with Linux fdisk and then making several ext2 systems on it works very well).
Now, if a fix is to be written, whatever the way it is written, it must recognise all previously partitioned disks (by Linux old fdisk as well as Win*), as well as the possible new common format. Is that really desirable ? I don't know.
-- Cyrille
------------------------------ Zog Zog
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu
| |