Re: Patch, forward release() return values to the close() call

[Oliver Neukum]

On Thursday 21 March 2002 09:27, Jeff Garzik wrote:
> Whoops, my apologies.  The patch looks ok to me.
>
> I read your text closely and the patch not close enough.  As I said, it
> is indeed wrong for a device driver to fail f_op->release(), "fail"
> being defined as leaving fd state lying around, assuming that the system
> will fail the fput().
>
> But your patch merely propagates a return value, not change behavior,
> which seems sane to me.

Hi,

close() does not directly map to release().
If you want your device to return error
information reliably, you need to implement flush().

	Regards
		Oliver