Re: [GIT PATCH] More PCI patches for 2.6.13

[Jeff Garzik <jgarzik@pobox.com>]

Linus Torvalds wrote:
> 
> On Sat, 10 Sep 2005, Jeff Garzik wrote:
> 
>>I -do- want to use iomap.  The problem is that no one has yet come up 
>>with a few that does all the proper resource reservation.  Everybody 
>>(including myself) did the ioread/iowrite part, but gave up before 
>>handling all cases of (a) legacy ISA iomap, (b) native PCI IDE iomap, 
>>and (c) non-standard MMIO iomap.
> 
> 
> It should all be trivial. The only ugly issue in the patch I just sent out 
> is that it needs to save the "legacy_mode" bits that were calculated at 
> initialization time somewhere in the ap structure. Then the 
> release_regions should match the request_regions.

More ugly issues abound, see below :)


> That's a cleanup, the current code is literally buggy. It may end up
> releasing IO address 0x1f0 twice, if somebody wasn't marked legacy, but
> actually had 0x1f0 in the PCI resource pointers (maybe that doesn't ever

Haven't run into anything yet that trips up the legacy/native detection 
in libata except for mixed mode (1 port legacy, 1 port native).  But 
those bugs aren't in the area of code we're discussing.


> happen, but still.. Relying on the legacy-value of the IO port instead of
> relying on whether you did a legacy request_region() is definitely at
> least conceptually wrong).

Its not that simple.  grep for ____request_region in both libata and the 
PCI quirks code.  libata grabs the SATA port on ICH boxes in combined 
mode... but has to do so before built-in IDE driver grabs them.

	Jeff