Re: [PATCH 1/2] streamline block SG_IO error processing

[Christoph Hellwig <hch@infradead.org>]

On Sat, Jan 15, 2005 at 01:35:32PM +1000, Douglas Gilbert wrote:
> I am working on the assumption that users of the
> SG_IO ioctl in the block layer (or via st or osst)
> want SCSI status and sense data returned
> via the ioctl immediately without:
>   - the error/warning silently disappearing and/or being
>     hidden by a retry (e.g. UNIT ATTENTION, lu becoming
>     ready)
>   - noise in the log (or console)
>   - any other side effects (save clearing expecting_cc_ua
>     if a UNIT ATTENTION was expected)

that's probably the right assumption.

> Changelog:
>   - cleanup scsi_end_request() documentation
>   - shorten path for block SG_IO through scsi_io_completion()
>   - for non-SG_IO sense processing in scsi_io_completion():
>      - ignore deferred errors (report + retry should suffice)
>      - consolidate into a cleaner switch statement

The patch looks extremly nice to me!

One extremly tiny nitpick though:

+	if (sense_valid && (! sense_deferred)) {

Both the space between ! and the variable and the superflous additional
bracket are against normal kernel style:

	if (sense_valid && !sense_deferred) {

But I think we can fix this while or after applying ;-)