Re: [SCSI] qla1280: endianess annotations

[Christoph Hellwig <hch@infradead.org>]

On Wed, Sep 07, 2005 at 10:13:21PM -0400, Jeff Garzik wrote:
> Linux Kernel Mailing List wrote:
> >diff --git a/drivers/scsi/qla1280.c b/drivers/scsi/qla1280.c
> >--- a/drivers/scsi/qla1280.c
> >+++ b/drivers/scsi/qla1280.c
> >@@ -1546,7 +1546,7 @@ qla1280_return_status(struct response * 
> > 	int host_status = DID_ERROR;
> > 	uint16_t comp_status = le16_to_cpu(sts->comp_status);
> > 	uint16_t state_flags = le16_to_cpu(sts->state_flags);
> >-	uint16_t residual_length = le16_to_cpu(sts->residual_length);
> >+	uint16_t residual_length = le32_to_cpu(sts->residual_length);
> > 	uint16_t scsi_status = le16_to_cpu(sts->scsi_status);
> [...]
> >+	__le16 status_flags;	/* Status flags. */
> >+	__le16 time;		/* Time. */
> >+	__le16 req_sense_length;/* Request sense data length. */
> >+	__le32 residual_length;	/* Residual transfer length. */
> >+	__le16 reserved[4];
> > 	uint8_t req_sense_data[32];	/* Request sense data. */
> 
> This isn't merely an endian annotation.
> 
> Is this a size fix, from 16 to 32, or a typo?  If its not a typo, 
> shouldn't the variable be declared 'uint32_t residual_length'?

It's a typo-fix.  The variable is 32bits in hardware.  Declaring it
in 32bit in software might make sense, but isn't that important.  Feel
free to send a patch if you care enough.