Re: sym53c875: reading /proc causes SCSI parity error

[David S. Miller]

   From: Gérard Roudier <[e-mail]>
   Date: Wed, 28 Nov 2001 20:51:01 +0100 (CET)
   
   On Wed, 28 Nov 2001, David S. Miller wrote:
   
   > Why not just put some bitmap pointer into the pci device
   > struct.  If it is non-NULL, it specifies PCI config space
   > areas which have side-effects.
   
   Or a simple offset beyond which reading data isn't desirable for
   whatever reasons.

I do not think that is sufficient.

I have seen chips where only one single PCI config space word would
trigger problems, and it was due to a hw bug.  The "offset beyond"
scheme would not allow to cover this case.