[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: ahd testing



On Wed, Jun 23, 2004 at 11:59:50PM -0500, Marco Peereboom wrote:

Hm. Regarding PCI-X, I suggest you:
	- try to test FreeBSD with the same HW
	- take a carefull look at the ahd_pci.c code regarding PCI-X, when I was porting ahd, I was forced to replace some macros with magic constants, like there:

static void
ahd_pci_split_intr(struct ahd_softc *ahd, u_int intstat)
...

/*
 * Check for splits in all modes.  Modes 0 and 1
 * additionally have SG engine splits to look at.
 */
 pcix_status = pci_conf_read(bd->pc, bd->tag,
        bd->pcix_off + 0x04);
			^^^^^

or for example there:

/*
 * Clear PCI-X status bits.
 */
 pci_conf_write(bd->pc, bd->tag, bd->pcix_off + 0x04,
       pcix_status);

That means, I think the PCI-X support needs to be improved in general..
The problem is, I really do not know, what PCI-X is, how it does function
and I really do not have the time at the moment..

Good luck towards stable ahd! :)

Milos

> krw@ and I have been working on ahd over the last few days at the hackathon and we are reasonably sure that ahd works very well as long as the older version of the chip is used.  The only issues we seem to have are whenever we use a chip version that is newer than the AIC7902A.  Our experience seems to indicate that the issues could be PCI-X related.  To narrow this down we would appreciate if people could give this a twirl and send us dmesg' plus success/failure stories.
> 
> Big thanks must go out to two people who made this development possible:
> * Diana Eichert went through quite some pain collecting donations to enable krw@ to purchase a board to develop with.
> * Milos Urbanek created the initial port from the FreeBSD code that was written by Justin T. Gibbs.  He did that in only a few short days and he did this without hardware!