From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alexander Viro Date: Sat, 26 Aug 2000 16:23:03 -0400 Message-Id: <200008262023.QAA01437@weyl.math.psu.edu> To: 9fans@cse.psu.edu Subject: Re: [9fans] New to Plan9 In-Reply-To: <200008231430.KAA27086@cse.psu.edu> Topicbox-Message-UUID: ffcb6792-eac8-11e9-9e20-41e7f4b1d025 In article <200008231430.KAA27086@cse.psu.edu> you write: >As for the 486 it sounds like more floppy problems >(something is amiss in the floppy driver, but no >one is sure what). I don't understand why having >an old floppy drive under Linux would mean you >can't mount disks of type vfat or ext2. I don't >know what tricks ext2 pulls, but vfat should be >using the same bits as fat does, i.e. the standard >1.44MB of data. ext2 is a garden variety FFS. It could not care less about the media or geometry - all accesses go through getblk() and bread() (or pagecache equivalents in 2.4) and it simply doesn't know anything about the underlying device. There are some _very_ sick tricks used by deranged DOS format(8) variants. Resulting diskette a) has different physical format on the first track and on the rest of the disk (the first track is shorter, usually 18 sectors vs. 30 or so) b) contains filesystem thinking that the first track has the same size as everything else (otherwise DOS would barf and die) c) relies on the fact that fatfs has reserved area it never touches. That reserved area is actually not present - it would live in the end of the first track if that track would have the same size as everything else. It works under DOS because DOS uses 3D addressing and never asks for the blocks in the end of first track. Linux floppy driver tries to cope with such horror when it meets one, but the best it can do is to refuse serving the requests for blocks in that "hole" - if it would just map the block 18 to the beginning of the second track fatfs would die, because it expects block 30 there. E.g. dd(1) will be very unhappy. IMO we shouldn't support that mess at all - yes, it's impossible to deal with from userland, but anyone dumb enough to use 2m and its ilk deserves to lose. That's the only case when ability to use the disk will depend on fs in question and it's _really_ not something you want to support. _If_ the diskettes with problems are in 2m format - just tell user to sod off or to do normal lowlevel format. Frankly, we should do the same, but maintainer of floppy driver has some strange ideas about being nice to lusers. It it's not a 2m perversion and Linux can do one fs type but not another - send the bug report. -- Fairy Tails start "Once upon a time." Army/Sea stories start "This is no shit." Software proposals start "1.0." Joe Zeff in the Monastery