In article <20010916145130.A260@rxgpc.iplus.net.tw>
rxg@ms25.url.com.tw wrote:

> I updated the driver with insteading of inb(), outb() by bus_space functions
> and remove the vtophys().
> So, it should be machine independent now.
> If someone has these cards and wants to test it, please fetch it from:
> 	http://members.tripodasia.com.tw/rxg/
> If we get no problem, I will send-pr. :)

Great!
It looks good, but I'd like some more cosmetics:

1) The PCI vendor ID and device ID are defined in sys/dev/pcidevs.h,
   so we should use them rather than define locally.

2) The driver now uses bus_space(9) properly, so it seems
   IOPortBase member of struct ACB is no longer needed.

3) All SRBs are allocated by bus_dmamem_alloc(), but
   it is better to use bus_dmamem_alloc() only for memory
   used to pass parameters to the adapter via DMA,
   i.e. only memory for SegmentX[] member should be
   allocated by bus_dmamem_alloc(). (see sys/dev/ic/iha*.[ch])
   Otherwise it causes some performance penalty because
   allocated memory by bus_dmamem_alloc() with BUS_DMA_COHERENT
   does not use cache on some architectures (like arm and mips).

4) bus_dmamap_sync with BUS_DMASYNC_PREWRITE is also needed
   for SegmentX[] before transfer.

5) It seems SRBSGPhyAddr member of struct SRB is not needed.
   It is refered only in trm_DataIO_transfer().

6) Maybe some byteswap code is required for big-endian machines.
   (But it should be trivial.)

7) It is much better to adapt these sources to KNF :-)
   (see /usr/share/misc/style)

(sorry I don't have the card.)
---
Izumi Tsutsui
tsutsui@ceres.dti.ne.jp