cvs commit: src/sys/compat/ndis kern_ndis.c subr_ndis.c

[Nate Lawson]

On Thu, 25 Mar 2004, M. Warner Losh wrote:
> In message: <20040325234024.GJ26131 at elvis.mu.org>
>             Alfred Perlstein <alfred at freebsd.org> writes:
> : * M. Warner Losh <imp at bsdimp.com> [040325 14:11] wrote:
> : > In message: <xzpvfksfztc.fsf at dwp.des.no>
> : >             des at des.no (Dag-Erling Sm?rgrav) writes:
> : > : Bill Paul <wpaul at FreeBSD.org> writes:
> : > :
> : > : des at dwp ~% gf BUS_GET_RESOURCE_LIST
> : > : qsrc/sys/compat/ndis/kern_ndis.c:       brl = BUS_GET_RESOURCE_LIST(device_get_parent(dev), dev);
> : > : src/sys/dev/if_ndis/if_ndis_pci.c:      rl = BUS_GET_RESOURCE_LIST(device_get_parent(dev), dev);
> : > : src/sys/kern/subr_bus.c:        rl = BUS_GET_RESOURCE_LIST(dev, child);
> : > : src/sys/kern/subr_bus.c:        rl = BUS_GET_RESOURCE_LIST(dev, child);
> : > : src/sys/kern/subr_bus.c:        rl = BUS_GET_RESOURCE_LIST(dev, child);
> : > : src/sys/kern/subr_bus.c:        rl = BUS_GET_RESOURCE_LIST(dev, child);
> : > : src/sys/kern/subr_bus.c:        rl = BUS_GET_RESOURCE_LIST(dev, child);
> : >
> : > Acutally, there are alot of downstream users of rl.
> :
> : :(
> :
> : Shouldn't we have macros:
> :
> : BUS_GET_RESOURCE_NEXT ?
> :
> : (instead of SLIST_NEXT)
>
> We already do.  bus_generic_rl_get_resource, et al.  The problem here
> is that the pci bus never intended its resources to be exposed outside
> of the bus layer.  But windows drivers need it, so there needs to be
> some tweaks need to be made.
>
> It looks like I was incorrect about the downstream uses of it.  I
> thought that pci and pccard did that, but it appears that they now use
> the common functions that matt dodd moved into subr_bus a while ago.

Fixing this to be FIFO instead of LIFO would also have helped for floppy
devices that specified a 6-port resource as three 2-port resources.  When
put in LIFO order, this gave 5-6,3-4,1-2 instead of 1-2,3-4,5-6.  The
latter would have been easier to parse.

-Nate