OLDCARD status

Project: OLDCARD upgrade to support PCI cards
Contat: imp@village.org
Funded by: Monzoon Networking, LLC

Monthly Status




The current patchset can be found at http://people.freebsd.org/~imp/pcic-stable.diff.16a the kernel part (relative to src/sys) and http://people.freebsd.org/~imp/pcic-stable.diff.16b the userland part (relative to src/usr.sbin/pccard/pccardd).


Old Monthly Status reports


This month has been a month of conventration and consolidation. Much of the changes from current have been migrating into stable. While this has been happening, bug reports have been coming in. There is one major unresolved issue, and several more minor ones.

This month I've added many things. I've improved the 3.3V support in older ISA bridge chipsets (which also is necessary in some of the newer PCI bridges). I've improved the ability of OLDCARD to cope with bridges that have been left in a non-default state by Windows and/or NEWCARD. I've also fixed many of the problems with suspending and resuming relating to pccard. I've also added support for disabling function 1 on pci chipsets that have been improperly manufactured.

Almost all of this functionality has been merged into stable. The pci attachment hasn't been merged due to the interrupt issue and the pci bios issue. The interrupt issue requires all people that use pccardd in current to add "-I -i X", where X is the interrupt of pcic. pccardd needs to be updated to cope with this automatically.

The PCI BIOS issue has two parts. The first part is that some machines do not support routing of interrupts via PCI BIOS due to their age. At least one pc98 laptop also suffers from this problem, as do many really old laptops that have first or second generation pci-pcmcia or pci-cardbus suffer from this. The PCI standard didn't add this until version 2.2. On these systems, interrupt routing will be difficult if not impossible to support. This means that many older PCI machines may not be able to support the wavelan pci bridges that lucent sells. The second problem is that some PCI BIOSes are buggy and panic the kernel when called.

The PCI interrupt routing code is incomplete in stable and may need to be specially enabled in order to function.

I'm racing the -stable deadline to get this merged into -stable to meet the 4.4 deadline.

$Id: oldcard-status.html,v 1.7 2001/08/04 20:51:36 imp Exp $