On 01/16/2011 10:12 PM, David Dillow wrote: > If the compiled object doesn't include linux/scatterlist.h before > scsi/scsi.h, it will get an incorrect definition of > SCSI_MAX_SG_CHAIN_SEGMENTS. > > Signed-off-by: David Dillow <dillowda@xxxxxxxx> > -- > This fixes an issue I found while working on the SRP initiator, where > linux/scatterlist.h was being pulled in by scsi/scsi_device.h. It's not > clear if I'm supposed to include linux/scatterlist.h myself, or if I'm > including the scsi headers in the wrong order. It makes sense for me > that scsi.h would pickup the needed headers itself to prevent confusion. > I totally agree. This below is the proper fix! > Another option is to potentially get rid of the ARCH_HAS_SG_CHAIN define > as all archs have it save m68knommu. > What's preventing m68knommu from chaining? this is pure C manipulations, what am I missing? Thanks for catching this. As is, it is vary dangerous. I wonder where else did it hit unnoticed? Boaz > > diff --git a/include/scsi/scsi.h b/include/scsi/scsi.h > index 648d233..b76d400 100644 > --- a/include/scsi/scsi.h > +++ b/include/scsi/scsi.h > @@ -9,6 +9,7 @@ > #define _SCSI_SCSI_H > > #include <linux/types.h> > +#include <linux/scatterlist.h> > > struct scsi_cmnd; > > -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html