Date: Sat, 21 Jan 2006 08:44:26 +0100 (CET) From: freebsdpcic@oldach.net (Helge Oldach) To: stable@freebsd.org Subject: OLDCARD vs. NEWCARD revisited Message-ID: <200601210744.k0L7iQnN020899@sep.oldach.net>
next in thread | raw e-mail | index | archive | help
Folks, I own an LG Electronics LW1100P wireless LAN PCI board, consisting of a Texas Instruments TI-1211 PCI-CardBus bridge and a wi(4)-compatible PCCARD. This card is recognized with an OLDCARD kernel but not with a(n almost) GENERIC kernel. I've spotted two oddities with this card and made it work with two rather trivial patches that essentially align NEWCARD a bit more to the OLDCARD behaviour: 1) The card (or the PCI/CardBus bridge?) seems to announce that it runs on 2V or 3.3V voltage. But it is actually lying about 2V support. It just doesn't work @ 2V - which is a rather pointless thing for a PCI card anyway. The delta between OLDCARD and NEWCARD is that OLDCARD did apply power starting with 5V, then 3.3V, then 2V, then 1V, but NEWCARD is doing it the other way 'round. Hence this card ends up with 2V supply and doesn't work. I've filed PR i386/91919 for this which contains a trivial patch reverting the voltage supply order. This might not be the preferred solution but it solves the issue. 2) The CIS on the card does not announce an IRQ preference at all. Hence the default IRQ mask of 0x0 applies and device_attach doesn't see an IRQ. But this is an I/O card which clearly needs an IRQ. I've filed PR i386/92070 for this which contains a trivial patch making 0xffff the default IRQ mask instead of 0x0. Maybe some kind soul might have a look at the PRs? Actually I think that some OLDCARD/NEWCARD issues that have been reported might be caused by these two oddities. BTW, The PR's relate to 5.4-STABLE but apply to 6.0-STABLE as well. Regards, Helge
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200601210744.k0L7iQnN020899>