> This note is sort of a half question half bug report.
[devices are accessed in blocksize-size units, and this unit is changed
upon a mount, and is 1024 to start with; this gives problems if one
wants to access the last few sectors of a drive]
Yes, a known problem.
It will go in 2.5.
The problem is old, but until recently nobody complained -
sacrificing 1 or 3 sectors at the end of a disk to the roundoff
was something like sacrificing 62 sectors to DOS-compatibility.
These days people put partition tables and other interesting stuff
at the very end of a disk and this suddenly becomes a problem.
If you really want to touch the last sector of a disk, then
you can start a partition an integral number of blocks before
the end. Messy, but at least all of the disk is accessible.
An ioctl to change the block size of a partition (or the full disk)
is slightly better, but is dangerous - it can be used only
when nothing is mounted, etc.
(My memory is bad, but it seems to me that such an ioctl was
proposed, and I suggested to add the checks [not in use,
no mount, not swap device] but I don't know what happened
to this proposal afterwards. Must check.)
Andries
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/