Messages in this thread | | | From | Bret Indrelee <> | Subject | RE: Specifying properly the PCI driver model on all linux archite | Date | Wed, 3 Nov 1999 14:30:25 -0600 |
| |
Alan Cox [mailto:alan@lxorguk.ukuu.org.uk] wrote: > > I think you will find that you'll have problems with your > hot swap and > > bridges, it really should be the PCI subsystem that decides > what areas of > > PCI are used or not. Once it is put in the configuration > register, that > > should be the end of it. > > > > I think that having all the PCI device drivers do > check_mem_region() and > > request_mem_region() is going to cause problems. > > The memory regions have three states so I think we are ok. They can be > > 1. Unused > 2. Allocated but not used (eg PCI but no driver loaded) > 3. Used > > As you clearly realise the major issue is getting type #1 > right so that > hot swap devices, AGP gart and the like can grab a hole
Actually, I was thinking the major issue was correctly handling alignment constraints on PCI-PCI Bridges (PPB) that may exist behind the hot-swap. You have to have the PPB addresses aligned on a 1MByte boundardary, and they can nest. You make a PPB hot-swappable, all of the sudden you have a bunch of cards which come in all at the same time behind the bridge and have to be allocated out of unused, contiguous space with a minimal address granularity of 1MBytes.
Is there anything preventing a PCI expansion unit from being hot-swappable? I haven't had a chance to read the hot-swap PCI specification, so I don't know off the top of my head. It should be possible.
-Bret
------------------------------------------------------------- SBS Technologies, Connectivity Products ... solutions for real-time connectivity
Bret Indrelee, Engineer SBS Technologies, Inc., Connectivity Products 1284 Corporate Center Drive, St. Paul MN 55121 Direct: (651) 905-4731 Main: (651) 905-4700 Fax: (651) 905-4701 E-mail: bindrelee@sbs-cp.com http://www.sbs.com -------------------------------------------------------------
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |