With recent dd and 2.4.17rc1 an attempt to write (with DD)
into a block-device yields EINVAL error.
Doing strace shows that it comes from ftruncate64()
on the opened device.
Should there be some other error code ? EBLKDEV ?
Or is the dd (4.0p) at fault, and it should either
not ask for the ftruncate*() at all, or ignore its error ?
/Matti Aarnio
On Mon, Dec 17, 2001 at 03:00:28AM +0200, Matti Aarnio wrote:
> With recent dd (4.0p) and 2.4.17rc1 an attempt to write (with DD)
> into a block-device yields EINVAL error.
With dd 4.1.3 the problem no longer exists.
That newer version does fstat() on file, and notices that
it is, indeed, a device.
End of problem.
But why I am seeing mysterious total hangs of this kernel is
another story, and probably related to memory allocation...
/Matti Aarnio