[ofa-general] Re: [PATCH] for-2.6.23 ib/umad: add partition support

[Roland Dreier]

 > Makes sense. If you like, an ioctl can be replaced with a write:
 > all 4-byte writes currently return -EINVAL.
 > 
 > This has a small advantage that write gets passed the buffer length
 > parameter, so it's easier to debug (e.g. strace outputs write buffers).

Hmm, I think I still prefer an ioctl to switch modes.  It makes for
cleaner separation of control and data path.

 > To make the interface more future-proof, we can 
 > ask all new-ABI users to use pwrite with offset 0,
 > and validate the offset in kernel.
 > Is this a good idea?

No, I don't like that interface.  Especially the converse interface of
pread() at offset 0 seems very confusing.

 - R.