RE: [PATCH 1/5] scsi: ufs: Allow UFS 3.0 as a valid version

[Jose Abreu]

From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Apr/25/2020, 12:10:56 (UTC+00:00)

> On Fri, Apr 24, 2020 at 01:36:56PM +0200, Jose Abreu wrote:
> > Add a define for UFS version 3.0 and do not print an error message upon
> > probe when using this version.
> 
> This doesn't really scale.  Version checks only make sense for a minimum
> supported version.  Rejecting newer versions is just a bad idea.
> 
> > @@ -8441,7 +8441,8 @@ int ufshcd_init(struct ufs_hba *hba, void __iomem *mmio_base, unsigned int irq)
> >  	if ((hba->ufs_version != UFSHCI_VERSION_10) &&
> >  	    (hba->ufs_version != UFSHCI_VERSION_11) &&
> >  	    (hba->ufs_version != UFSHCI_VERSION_20) &&
> > -	    (hba->ufs_version != UFSHCI_VERSION_21))
> > +	    (hba->ufs_version != UFSHCI_VERSION_21) &&
> > +	    (hba->ufs_version != UFSHCI_VERSION_30))
> 
> i.e. this should become
> 
> 	if (hba->ufs_version < UFSHCI_VERSION_10)
> 
> as an additional cleanup I think it makes more sense t use a UFSHCI_VER()
> macro similar to KERNEL_VERSION() or NVME_VS() instead of adding a new
> define for every version.

Yeah, unfortunately I don't think this can be done because of this:

enum {
	UFSHCI_VERSION_10 = 0x00010000, /* 1.0 */
	UFSHCI_VERSION_11 = 0x00010100, /* 1.1 */
	UFSHCI_VERSION_20 = 0x00000200, /* 2.0 */
	UFSHCI_VERSION_21 = 0x00000210, /* 2.1 */
};

So, version 1.0 and 1.1 have higher values of 2.0 and 2.1 in terms of 
absolute value.

---
Thanks,
Jose Miguel Abreu