Re: [PATCH 1/11] qla2xxx: Add ISP24xx definitions.

[Christoph Hellwig <hch@infradead.org>]

> diff --git a/drivers/scsi/qla2xxx/qla_def.h b/drivers/scsi/qla2xxx/qla_def.h
> --- a/drivers/scsi/qla2xxx/qla_def.h
> +++ b/drivers/scsi/qla2xxx/qla_def.h
> @@ -62,6 +62,22 @@
>  #define PCI_DEVICE_ID_QLOGIC_ISP6322	0x6322
>  #endif
>  
> +#ifndef PCI_DEVICE_ID_QLOGIC_ISP2422
> +#define PCI_DEVICE_ID_QLOGIC_ISP2422	0x2422
> +#endif
> +
> +#ifndef PCI_DEVICE_ID_QLOGIC_ISP2432
> +#define PCI_DEVICE_ID_QLOGIC_ISP2432	0x2432
> +#endif
> +
> +#ifndef PCI_DEVICE_ID_QLOGIC_ISP2512
> +#define PCI_DEVICE_ID_QLOGIC_ISP2512	0x2512
> +#endif
> +
> +#ifndef PCI_DEVICE_ID_QLOGIC_ISP2522
> +#define PCI_DEVICE_ID_QLOGIC_ISP2522	0x2522
> +#endif

It's only an historic accident that the older ids are in here, care
to remove all and move them to pci_ids.h instead?

> +	int		(*start_scsi)(srb_t *);

umm, that's most of ->queuecommand, isn't it?  Shouldn't you define
a separate scsi_host_template for 24xx/25xx?

> +//ISP24xx

please try to avoid C++-style comments

> +	volatile uint16_t mailbox0;
> +	volatile uint16_t mailbox1;
> +	volatile uint16_t mailbox2;

never use volatile, always use proper kernel primitives to access
hardware or atomic_t for atomic in-kernel variables.