On Tue, 3 Apr 2018, Christoph Hellwig wrote:
>
> > +
> > + if (zep->zorro3) {
> > + /* Only Fastlane Z3 for now - add switch for correct struct
> > + * dma_registers size if adding any more
> > + */
> > + esp->dma_regs = ioremap_nocache(dmaaddr,
> > + sizeof(struct fastlane_dma_registers));
> > + } else
> > + esp->dma_regs = (void __iomem *)ZTWO_VADDR(dmaaddr);
>
> doesn't this need a __force (and a comment on why it is safe) to make
> sparse happy?
>
Sparse is happy with that cast. These are the warnings I get:
CHECK /home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:976:33: warning: incorrect type in assignment (different address spaces)
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:976:33: expected unsigned long *board_base
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:976:33: got void [noderef] <asn:2>*
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1068:28: warning: incorrect type in argument 1 (different address spaces)
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1068:28: expected void [noderef] <asn:2>*addr
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1068:28: got unsigned long *board_base
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1096:28: warning: incorrect type in argument 1 (different address spaces)
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1096:28: expected void [noderef] <asn:2>*addr
/home/fthain/src/kernel.org/linux/drivers/scsi/zorro_esp.c:1096:28: got unsigned long *board_base
which seems to indicate that board_base should have type void __iomem *.
--
